JP4055703B2 - Robot hand device - Google Patents

Description

  The present invention relates to a robot hand apparatus that can place a grasped object without giving an impact to the object.

Robot hand devices have been used in various fields as various industrial robots and humanoid robots. The robot hand apparatus is mainly used to grip an object having a predetermined shape from a predetermined direction. In recent years, a robot hand device having a multi-finger robot hand has been developed, and an object of any shape can be gripped from any direction by a multi-finger robot hand having multiple degrees of freedom (Patent Literature). 1).
JP 2001-277175 A JP 2001-310284 A

  However, when an object of an arbitrary shape is gripped or an object is gripped from an arbitrary direction, it is very difficult to determine the timing of separating the object from the multi-fingered robot hand when placing the object on the floor or the like. Therefore, if the timing is too early, the object will fall to the floor, or if the timing is late, the object will be pressed against the floor, which may cause an impact on the object and damage the object. There is.

  Accordingly, an object of the present invention is to provide a robot hand device that can place an object without giving an impact to the object.

  The robot hand device according to the present invention performs control for separating an object from the robot hand when a reaction force received from a place where the object is placed exceeds a threshold when placing an object grasped by the robot hand, and sets the threshold to the weight of the object. The value is proportional to.

  In this robot hand device, when placing a grasped object, it is determined whether or not the reaction force received by the object exceeds a threshold value, and control is performed to separate the object from the robot hand when the threshold value is exceeded. In this way, in this robot hand device, the timing for releasing the object is determined based on the reaction force received by the object, so the optimum timing for placing the object is determined regardless of the shape of the object and the direction in which the object is gripped. can do. For this reason, the object is not pushed against the place where the object is dropped because the timing is too early or the object is placed while the object is grasped while the timing is too late, and the object can be placed without giving an impact to the object.

  The robot hand device of the present invention may be configured to calculate the reaction force from the difference between the weight of the object and the force for gripping the object in the robot hand.

  In this robot hand device, the reaction force is obtained by calculation from the difference between the weight of the object and the force grasped by the robot hand, and control is performed to release the object when the reaction force obtained by this calculation exceeds a threshold value. Therefore, this robot hand device can easily determine the reaction force, and can determine the release timing without means for actually detecting the reaction force.

In the above robot hand device of the present invention, the fingertip resultant force in the robot hand coordinate system is converted into the fingertip resultant force in the robot hand device coordinate system according to the attitude of the robot hand from the robot hand device base coordinate system to the robot hand coordinate system. The weight of the object may be estimated based on the vertical fingertip resultant force in the fingertip resultant force in the base coordinate system of the robot hand device .

  In this robot hand device, since the weight of the object is estimated based on the gripping force of the robot hand and the posture of the robot hand, the weight of the object can be detected regardless of the shape of the object and the direction in which the object is gripped. Can do. Therefore, this robot hand apparatus does not need to measure the weight in advance before grasping the weight of the object, or to include means for actually detecting the weight of the object.

  In the robot hand device of the present invention, it is preferable that the threshold value is a value proportional to the weight of the object.

  In this robot hand apparatus, the reaction force increases as the weight of the object increases. Therefore, by setting the threshold value to be a value proportional to the weight of the object, the separation timing can be determined with high accuracy according to the weight of the object.

  According to the present invention, a grasped object can be placed without giving an impact to the object.

  Hereinafter, an embodiment of a robot hand apparatus according to the present invention will be described with reference to the drawings.

  In the present embodiment, the robot hand device according to the present embodiment is applied to a robot hand device that grips an object having an arbitrary shape and an arbitrary weight and places the gripped object on a predetermined floor surface. In the robot hand device according to the present embodiment, the robot hand is attached to the tip of the arm, and the robot hand has four fingers.

  A configuration of the robot hand apparatus 1 will be described with reference to FIGS. 1 and 2. FIG. 1 is a configuration diagram of the robot hand apparatus according to the present embodiment. FIG. 2 is a structural diagram of the robot hand of the robot hand device according to the present embodiment.

  The robot hand apparatus 1 grips an object whose weight and shape are unknown with a fingertip, and places the gripped object on a predetermined floor surface. In particular, in the robot hand apparatus 1, the weight of the object is estimated, and the estimated value of the weight can be used to place the object quietly on the floor surface. The robot hand device 1 mainly includes an arm 2, a robot hand 3, a camera 4, a rotary encoder 5, a six-axis force sensor 6, and a control device 7.

  The arm 2 includes a plurality of joints and links connecting the joints. Each joint is a rotary joint and is constituted by a motor 21. The motor 21 is supplied with a driving current DI from the motor driver 20 and is driven to rotate at a predetermined angle in a predetermined direction in accordance with the driving current DI. The motor driver 20 receives a command signal CS from the control device 7 and generates a drive current DI according to the rotation direction and angle commanded by the command signal CS, or according to a stop command of the command signal CS. The generation of the drive current DI is stopped.

  The robot hand 3 is composed of four fingers: a thumb 3a, an indicating finger 3b, a middle finger 3c, and a ring finger 3d (see FIG. 2). The thumb 3a includes four motors 31,... With 4 degrees of freedom. The other three fingers 3b to 3d have three motors 31,. Further, since the three fingers 3b to 3d include the interlocking joint 32, the number of joints is four. The degree of freedom of the robot hand 3 as a whole is 13. Similarly to the motor 21, the motor 31 is rotationally driven by the drive current DI from the motor driver 30. Similar to the motor driver 21, the motor driver 30 generates / stops the generation of the drive current DI according to the command signal CS from the control device 7. The fingertip portions of the fingers 3a to 3d are curved elastic bodies. In FIG. 2, only three motors 31, 31, 31 are drawn for the thumb 3 a and only two motors 31, 31 are drawn for the other fingers 3 b to 3 d, but there are motors behind the six-axis force sensor 6. Another one is provided.

  In order to recognize the position, orientation, shape, and the like of the object, the camera 4 captures the object before gripping and transmits the image data to the control device 7 as an image signal PS.

  The rotary encoder 5 is provided at each joint of the arm 2 and detects the rotation angle of each joint. Then, the rotary encoder 5 transmits the detected rotation angle as a joint angle signal AS to the control device 7. Incidentally, the joint angle may be detected by a potentiometer or the like instead of the rotary encoder.

  The six-axis force sensor 6 is provided at the base of each finger 3a to 3d of the robot hand 3 (see FIG. 2), and detects three components of force in each axial direction and three components of moment about each axis. . Then, the six-axis force sensor 6 transmits a six-axis force signal FS composed of the detected six components to the control device 7.

  The control device 7 will be described with reference to FIGS. FIG. 3 is a diagram showing a robot coordinate system and a hand coordinate system. 4A and 4B are explanatory views of the posture of the robot hand, where FIG. 4A is an example showing the configuration of the degree of freedom of the arm, and FIG. 4B is a link coordinate system at each joint of the arm shown in FIG. It is a figure which shows the relationship between a coordinate system and a hand coordinate system. FIG. 5 is a diagram illustrating a temporal change in the fingertip resultant force in the vertical direction. FIG. 6 is a diagram illustrating the relationship between the estimated object weight value and the floor reaction force threshold value. FIG. 7 is a diagram illustrating a temporal change in the fingertip resultant force in the vertical direction when placing an object.

  The control device 7 is an electronic control unit including a CPU [Central Processing Unit], a ROM [Read Only Memory], a RAM [Random Access Memory], and the like. The control device 7 takes in the detection signals PS, AS, FS from the detection means 4, 5, 6 to hold the target object and place it on a predetermined floor surface. The command signal CS is transmitted to 30,... To control the driving of the motors 21,. In particular, the control device 7 performs object weight estimation control for estimating the weight of the object and control for placing an object to be placed on the floor without giving a shock to the object. The object weight estimation control and the object placement control will be described in detail.

Before describing each control, the coordinate system used in the control device 7 will be described. As shown in FIG. 3, the object O is grasped by the fingertips of the four fingers 3 a to 3 d and lifted by the arm 2, and the robot hand 3 is in an arbitrary posture with respect to the base of the robot hand apparatus 1. The coordinate system of the robot hand 3 is defined as a hand coordinate system Σh (X _h , Y _h , Z _h ) ^T, and the base coordinate system of the robot hand device 1 is defined as the robot coordinate system ΣR (X _R , Y _R , Z _R ). Define ^T. In this hand coordinate system Σh, the resultant force (F _x ^h , F _y ^h , F _z ^h ) ^T of each fingertip force of the four fingers 3 a to 3 d when holding the object O is defined.

The link coordinate system of the arm 2 will be described with reference to the example of the arm 2 composed of seven joints (motors 21) shown in FIG. As shown in FIG. 4A, the arm 2 has seven degrees of freedom. As shown in FIG. 4B, a reference coordinate system Σ0 (X ₀ , Y ₀ , Z ₀ ) ^{T of} the arm 2 is defined with respect to the robot coordinate system ΣR. A link coordinate system Σ1 (X ₁ , Y ₁ , Z ₁ ) ^{T to} Σ ₇ (X ₇ , Y ₇ , Z ₇ ) ^T is defined for each joint, and the Z axis of the link coordinate system Σ _{1 to} Σ ₇ is the rotation axis. Is set to be The terminal coordinate system ΣE (X _E , Y _E , Z _E ) ^T is the hand coordinate system Σh. In the arm 2, the rotation matrix T _{i−1, i} from the link coordinate system Σi−1 to the next link coordinate system Σi is expressed by the equation (1) when only the rotational motion is considered without considering the translational movement. Can do. Furthermore, the rotation matrix TR _{, h} from the robot coordinate system ΣR to the hand coordinate system Σh can be expressed by Equation (2).

In equation (1), φ _i-1 is rotation around the X _i-1 axis, and is an angle (constant) determined between the link coordinate system Σi-1 and the link coordinate system Σi. θ _i is a rotation angle around the Z _i-1 axis (that is, the Z _i axis) after rotation, and is a joint angle detected by the rotary encoder 5. Therefore, the rotation matrix TR _{, h} can be obtained from the joint angles at each joint. The rotation matrix TR _{, h} indicates the posture (hand posture) of the robot hand 3.

  The object weight estimation control in the control device 7 will be described. By knowing the weight of the object, it is possible to set an optimum gripping force for gripping the object by the robot hand 3 and to determine the timing for releasing the object when placing the object. Therefore, in the object weight estimation control, the weight of the object is estimated with high accuracy based on the fingertip resultant force and the hand posture.

In the control device 7, the rotation matrix TR _{, h} is obtained by Expression (2) using each joint angle by the joint angle signal AS. Further, the control device 7 obtains the fingertip resultant force (F _x ^h , F _y ^h , F _z ^h ) ^T in the hand coordinate system Σh using the six components of each finger by the six-axis force signal FS. Then, the control device 7 obtains the fingertip resultant force (F _x ^R , F _y ^R , F _z ^R ) ^T in the robot coordinate system ΣR by Expression (3).

_F ^{z R} fingertip force in the robot coordinate system ^{_{ΣR (F x R, F y}} R, F z R) in ^T is the vertical direction of the fingertip force. Theoretically, the weight W _O of the object can be obtained by Expression (4) using the fingertip resultant force F _z ^R in the vertical direction.

W _f in equation (4) is the weight of the entire finger (finger weight) and is known.

Generally, in order to eliminate the influence of sensor drift in a no-load state before gripping, it is necessary to perform the offset removal processing of the six-axis force sensor 6 once. The output value of the 6-axis sensor 6 immediately after this offset processing becomes 0, and the component force in each axial direction of the 6-axis force sensor 6 due to the self-weight of the finger also becomes 0, and is cancelled. Therefore, when the hand posture (rotation matrix T _{R, h} ) changes due to the movement of the arm 2, the component force in each axial direction of the six-axis force sensor 6 due to the weight of the finger itself also changes. That is, the fingertip resultant force F _z ^{R in the} vertical direction changes depending on the hand posture. Therefore, in order to estimate the weight of an object in an arbitrary hand posture, the influence of this finger weight must be corrected.

Therefore, in the control device 7, when the offset removal of the six-axis force sensor 6 is performed, the components in the respective axial directions of the canceled finger weight are obtained. Specifically, assuming that the rotation matrix indicating the hand posture at that time is T ′ _{R, h} , each axial direction component (W _x ^R , W _y ^R , W _z ^R ) in the robot coordinate system ΣR of the finger weight W _f. ^T and the sensor coordinate system (i.e., the hand coordinate system .SIGMA.H) each axial component of ^{_{(W x h, W y h}} , W z h) relationship with ^T can be expressed by equation (5). Furthermore, from equation (5), each axial direction component (W _x ^h , W _y ^h , W _z ^h ) in the sensor coordinate system can be expressed by equation (6).

The control device 7 takes in the joint angle signal AS at the time of offset processing from the rotary encoder 5 and obtains a rotation matrix T ′ _{R, h} that is the hand posture at the time of offset processing using the joint angle. Then, the control device 7 uses the components T ₃₁ , T ₃₂ , T ₃₃ of the rotation matrix T ′ _{R, h} , and the axial components (W _x ^h , W _y ) in which the finger weight is canceled from the equation (6). ^h , W _z ^h ) ^T is obtained. When the influence of the finger weight is corrected using the respective axial direction components (W _x ^h , W _y ^h , W _z ^h ) ^T with respect to Expression (3), Expression (7) is obtained.

In Expression (7), the rotation matrix T ″ _{R, h} is an arbitrary hand posture after the offset processing, and (F _x ^h , F _y ^h , F _z ^h ) ^T is the fingertip in the hand coordinate system Σh after the offset processing. (F _x ^R , F _y ^R , F _z ^R ) ^T is a fingertip resultant force in the robot coordinate system ΣR after the offset process In the control device 7, the joint angle signal AS after the offset process is output from the rotary encoder 5. And the rotation matrix T ″ _{R, h} after the offset processing is obtained using the joint angle. Further, the control device 7 takes in the 6-axis force signal FS after the offset processing from the 6-axis force sensor 6 and applies the fingertip resultant force (F _x ^h , F _y ^h) in the hand coordinate system Σh after the offset processing from the 6 components of each finger. , F _z ^h ) ^T is obtained. Then, the control device 7 obtains the fingertip resultant force (F _x ^R , F _y ^R , F _z ^R ) ^T in the robot coordinate system ΣR after the offset process, according to the equation (7).

The fingertip resultant force F _z ^R in the vertical direction obtained by the equation (7) includes a component of the finger weight. In the robot coordinate system .SIGMA.R, since its own weight of the finger is always along the negative direction of the Z _R-axis, the estimated value W _O of the object weight can be determined by Equation (8).

In equation (8), F _z ^R is always a negative value, and W _O and W _f are always positive values. In the control device 7, the estimated value W _O of the object weight is obtained from the equation (8) using the vertical fingertip resultant force F _z ^R obtained from the equation (7). In fact, the value of W _O is changed at each sampling by the influence of noise six-axis force sensor 6 and the rotary encoder 5, a predetermined time (e.g., 1 second) the W _O obtained in averaged, the object and the average value Used as an estimate of weight. Incidentally, even when the 6-axis force sensor is provided at a position other than the base of the finger, the weight of the portion ahead of the 6-axis force sensor can be canceled by the same method as described above.

  Next, control for placing an object in the control device 7 will be described. When placing an object, it is important at what timing the robot hand 3 releases the object. For example, if you release it too early, objects can fall to the floor and be damaged. On the other hand, if the release is too slow, the object continues to be placed even after the object hits the floor, so that the object may be damaged by the force pressing the object against the floor. Therefore, in order to place an object on the floor without giving an impact, it is necessary to determine whether or not the object hits the floor. Therefore, in the control of placing an object, the reaction force from the floor is obtained in real time based on the estimated value of the weight of the object, and the separation timing is determined with high accuracy by monitoring the floor reaction force.

When the operation of placing an object is started, the control device 7 takes in the joint angle signal AS from the rotary encoder 5 and the six-axis force signal FS from the six-axis force sensor 6 and calculates the robot coordinate system ΣR in real time according to Equation (7). The fingertip resultant force F _z ^R in the vertical direction at is determined. Since the floor reaction force is generated when the object comes into contact with the floor, the floor reaction force is obtained by using the vertical fingertip resultant force F _z ^R obtained in real time. As shown in Expression (9), the floor reaction force F is defined by the difference between the estimated weight W _O of the object and the fingertip resultant force F _z ^R in the vertical direction at a certain time.

In Expression (9), the downward direction in the vertical direction is the positive direction. In the control device 7, the floor reaction force F is obtained in real time according to Equation (9) based on the estimated object weight W _O obtained in advance and the fingertip resultant force F _z ^R in the vertical direction obtained in real time.

Theoretically, when the floor reaction force F is 0, the object is not in contact with the floor, and when the floor reaction force F is greater than 0, it can be determined that the object is in contact with the floor. However, actually, as shown in FIG. 5, the fingertip resultant force F _z ^{R in} the vertical direction obtained by the control device 7 varies due to the influence of noise in the sensor. Therefore, the fingertip resultant force F _z ^{R in the} vertical direction fluctuates around the estimated value W _O of the weight of the object at the boundary of whether or not the object contacts the floor. Therefore, even if the floor reaction force F is greater than 0, the object is not necessarily in contact with the floor.

The control device 7 sets an optimum floor reaction force threshold value in order to prevent erroneous determination of the timing at which an object is released due to the influence of noise of the sensor. If this floor reaction force threshold is a constant value, it cannot be applied to all objects having different weights. Further, when the floor reaction force threshold value is set in proportion to the weight of the object, if the weight of the object is too light, the threshold value changes the fluctuation range A of the fingertip resultant force F _z ^{R in} the vertical direction (the difference between the maximum value and the minimum value). ) (See FIG. 5). Therefore, when the object is light, the floor reaction force threshold is set according to the fluctuation range A so that it does not become too small. On the other hand, when the object is heavy, the floor reaction force increases according to the weight of the object, and is set according to the weight of the object. Specifically, as shown in Expression (10), the floor reaction force threshold is set to a value obtained by multiplying the fluctuation range A by α when the estimated value W _O of the object weight is smaller than αA / β, When the estimated value W _O is greater than or equal to αA / β, the estimated value W _O of the object weight is multiplied by β.

6, the formula (10) shows the relationship between the object weight estimates W _O and the floor reaction force threshold by the threshold, the W _O is constant value is .alpha.A / beta smaller area, W _O Increases in proportion to W 2 _O in the region of αA / β or more.

In order to set the floor reaction force threshold, the control device 7 monitors the fingertip resultant force F _z ^R in the vertical direction in real time for 1 second, extracts the maximum value and the minimum value in that 1 second, and determines the maximum value. And the difference A (variation range) between the minimum value and the minimum value. Then, the control device 7 sets the floor reaction force threshold value by the equation (10) based on the fluctuation range A and the estimated object weight value W _O obtained in advance. The reason for monitoring for 1 second is to detect the fluctuation range A with as high accuracy as possible, considering that the placing operation takes 1 second or more.

And in the control apparatus 7, the floor reaction force F calculated | required in real time and a floor reaction force threshold value are compared. When the floor reaction force F is less than or equal to the floor reaction force threshold value, the control device 7 continues to drive the arm 2 and keep the object placed at a constant speed. When the floor reaction force F exceeds the floor reaction force threshold, the control device 7 ends the placing operation. When ending the placing operation, the control device 7 transmits a command signal CS for stopping the arm 2 (stops the motors 21 of each joint) to the motor driver 20. A command signal CS for releasing the gripping of the object by the hand 3 (rotating drive of each motor 31,...) Is transmitted to the motor driver 30. As shown in FIG. 7, when the object contacts the floor, the fingertip resultant force F _z ^{R in the} extending direction decreases, so the floor reaction force F increases and the floor reaction force F exceeds the floor reaction force threshold. At that time, the arm 2 is stopped, and immediately after the arm 2 stops, the gripping of the robot hand 3 is released.

  With reference to FIGS. 1-2, the operation | movement in the robot hand apparatus 1 is demonstrated. First, a series of operations from gripping an object in the robot hand device 1 to placing the object will be described with reference to the flowchart of FIG. Further, the object weight estimation control in the control device 7 will be described with reference to the flowchart of FIG. 9, and the control of placing an object will be described with reference to the flowchart of FIG. FIG. 8 is a flowchart showing the operation of the robot hand apparatus of FIG. FIG. 9 is a flowchart showing object weight estimation control in the control apparatus of FIG. FIG. 10 is a flowchart showing control for placing an object in the control apparatus of FIG.

  First, a series of operations in the robot hand apparatus 1 will be described. The control device 7 takes in the image signal PS from the camera 4 and recognizes the position, orientation, shape, etc. of the object to be grasped by image recognition (S1). Then, the control device 7 determines the object gripping method (gripping posture or target gripping force) from the direction and shape of the object, and further determines the target gripping position from the position of the object (S2). Subsequently, the control device 7 calculates the target posture of each finger 3a to 3d and the target posture of the arm 2 (target joint angle at each joint) by inverse motion analysis (S3).

  Then, the control device 7 creates a time-series joint angle command for each joint according to the target joint angle, and transmits a command signal CS corresponding to the joint angle command to the motor driver 20 to drive the motor 21 (S4). ). Then, the arm 2 drives and moves the robot hand 3 to a position for gripping an object. Immediately before the gripping, the control device 7 temporarily stops the arm 2 and performs the offset process of the six-axis force sensor 6. After the offset process, the control device 7 transmits a command signal CS for moving the robot hand 3 to the target gripping position for each joint, and drives the motor 21 (S4). At the same time, the control device 7 transmits a command signal CS for gripping an object with a target gripping force to the motor driver 30 for each finger to drive the motor 31 (S4). Then, the robot hand 3 is driven and grips an object (S4).

  Subsequently, the control device 7 continues the grip control by the robot hand 3 and transmits a command signal CS for lifting the arm 2 to the motor driver 20 to drive the motor 21 (S5). Then, the arm 2 drives and lifts the object gripped by the robot hand 3 (S5).

  Next, the control device 7 continues the grip control by the robot hand 3 and transmits a command signal CS for stopping the arm 2 to the motor driver 20 to stop driving the motor 21 (S6). Then, the arm 2 stops and stops the object at a predetermined position (S6). During this stop, the control device 7 estimates the object weight (S7).

  When the estimation is completed, the control device 7 transmits a command signal CS for moving the object above a predetermined place where the object is placed to the motor driver 20 to drive the motor 21 (S8). Then, the arm 2 is driven to move the object to above the place where it is placed (S8). Further, the control device 7 transmits a command signal CS for moving the object downward at a constant speed to the motor driver 20 to drive the motor 21 (S9). Then, the arm 2 starts the placing operation, and moves the object at a constant speed along the vertical direction (S9).

  When the placing operation is started, the control device 7 performs control for placing the object so as not to give an impact to the object when placing, and determines the timing for releasing the object (S10). When the control device 7 determines the timing of releasing the object, the control device 7 transmits a command signal CS for stopping the arm 2 to the motor driver 20 to stop driving the motor 21 (S11). Then, the arm 2 stops (S11). Subsequently, in the control device 7, for each finger, a command signal CS for releasing the gripping of the object is transmitted to the motor driver 30 to drive the motor 31 (S11). Then, the robot hand 3 releases the object and places the object on the floor (S11).

The object weight estimation control in the control device 7 will be described. At the time of offset processing of the six-axis force sensor 6, the control device 7 reads joint angle signals AS from the rotary encoders 5 at the respective joints _, and calculates a rotation matrix T ′ _{R, h} that is a hand posture at the time of offset processing (S20). ). Further, the control device 7 uses the components T ₃₁ , T ₃₂ , and T ₃₃ of the rotation matrix T ′ _{R, h and} the axial components (W _x ^h , W) of the finger weight canceled by Equation (6). _y ^h, to calculate the ^{_{W z h) T (S20)}} .

During the processing of S7, the control device 7 reads the 6-axis force signals FS from the 6-axis force sensors 6 of the four fingers 3a to 3d, respectively, and applies the fingertip resultant force (F _x ^h , F _y ^h in the hand coordinate system Σh). , F _z ^h ) ^T is calculated (S21). Subsequently, the control device 7 reads the joint angle signal AS from the rotary encoder 5 in each joint _, and calculates the rotation matrix T ″ _{R, h} which is the hand posture after the offset processing (S22).

Then, in the control device 7, each axial component (W _x ^h , W _y ^h , W _z ^h ) ^{T of} the canceled finger weight, and the fingertip resultant force (F _x ^h , F _y ^h , F _z ) in the hand coordinate system Σh. ^h ) Based on the hand posture after the offset processing (rotation matrix T ″ _{R, h} ), the fingertip resultant force F _z ^R in the vertical direction in which the influence of the finger weight is corrected by the equation (7) is calculated (S23).

Finally, in the control device 7, the weight W _O of the object is estimated by Expression (8) using the fingertip resultant force F _z ^R in the vertical direction and the finger weight W _f (S24).

Control of placing an object in the control device 7 will be described. When the operation of placing an object is started, the control device 7 reads the 6-axis force signals FS from the 6-axis force sensors 6 of the four fingers 3a to 3d at regular time intervals, and the fingertips in the vertical direction according to the equation (7). The resultant force F _z ^R is calculated. Then, the control unit 7 detects the maximum value and the minimum value in the vertical direction of the fingertip force F _z ^R in one second, to set the variation range A (S30). Further, the control device 7 determines the floor reaction force threshold value by the equation (10) based on the fluctuation range A and the estimated value W _O of the object weight (S31).

Subsequently, the control device 7, at regular time intervals, the six-axis force sensor 6 four fingers 3a~3d read each 6-axis force signal FS, fingertip force of the current vertical direction by Equation (7) F _z ^R is calculated (S32). Further, the control device 7 calculates the floor reaction force F by Equation (9) based on the current fingertip resultant force F _z ^{R in} the vertical direction and the estimated value W _O of the weight of the object (S33). Then, the control device 7 determines whether or not the floor reaction force F obtained by the calculation has become larger than the floor reaction force threshold value (S34). In this way, the control device 7 continues to detect the floor reaction force in real time during the operation of placing the object, and determines the timing of releasing the object depending on whether the floor reaction force exceeds the floor reaction force threshold value.

  While the floor reaction force F is equal to or less than the floor reaction force threshold, the control device 7 continues the control of placing the object at a constant speed by the arm 2 (S35), and returns to the process of S32. In this case, since the floor reaction force F does not indicate that the object has contacted the floor, the floor reaction force F is repeatedly obtained to determine the change in the floor reaction force F.

  When the floor reaction force F becomes larger than the floor reaction force threshold, the control device 7 immediately stops the driving of the arm 2 and performs control for releasing the gripping of the object by the robot hand 3 (S36). The object is then placed gently on the floor. In this case, since the floor reaction force F indicates that the object has contacted the floor, the operation of placing the object is immediately terminated.

  According to this robot hand device 1, the object is released regardless of the shape of the object or the direction in which the object is gripped by determining whether or not the floor reaction force obtained in real time during the operation of placing the object exceeds the threshold value. The optimal timing can be determined. Therefore, since the object is not released quickly and dropped on the floor, or the object is released slowly and pressed against the floor, it can be placed quietly without giving an impact to the object. Furthermore, in the robot hand apparatus 1, the floor reaction force is obtained by a simple calculation from the weight of the object and the fingertip resultant force in the vertical direction, so that no means for detecting the floor reaction force is required. In the robot hand apparatus 1, the floor reaction force threshold value is set based on the fluctuation range of the object weight and the vertical fingertip resultant force, so that the timing of releasing the object due to the influence of sensor noise or the object weight is incorrect. There is no determination, and the determination accuracy of the release timing is high.

  Furthermore, according to the robot hand apparatus 1, by estimating the weight of the object based on the fingertip resultant force and the hand posture, it is possible to know the weight of the object regardless of the shape and orientation of the object and the direction in which it is gripped. Therefore, the robot hand apparatus 1 does not require means for detecting the weight of the object in advance or directly detecting the weight of the object. In particular, the robot hand device 1 corrects the influence of the finger's own weight when determining the weight of the object, so that the weight of the object can be estimated with high accuracy.

  As mentioned above, although embodiment which concerns on this invention was described, this invention is implemented in various forms, without being limited to the said embodiment.

  For example, although the present embodiment is applied to a multi-fingered robot hand having four fingers, the present invention can also be applied to other robot hands such as a multi-fingered robot hand other than four fingers and a robot hand having a palm. In this embodiment, the present invention is applied to a robot hand having a total of 13 degrees of freedom. However, the degree of freedom of each finger can be applied to any degree of freedom.

  In this embodiment, the present invention is applied to gripping an object with a fingertip. However, the present embodiment can also be applied to gripping an object with the surface of a hand such as a palm. In this case, a six-axis force sensor is provided on the wrist. The object weight is estimated and the floor reaction force is calculated using the force of grasping the object with the hand surface instead of the fingertip resultant force.

  Further, in the present embodiment, an object with an unknown weight is targeted and the weight of the object is estimated, but when an object with an already known weight is targeted, the weight of the object is obtained by another method or Even when the weight of the object is directly detected, the floor reaction force may be calculated from the object weight obtained by another method. In this embodiment, the floor reaction force is obtained by calculation. However, even when the floor reaction force is acquired by another method such as directly detecting the floor reaction force, the floor reaction force acquired by another method is used. The timing of releasing may be determined. In this embodiment, the threshold is set according to the fluctuation range of the fingertip resultant force in the vertical direction and the weight of the object. However, if the object is a light object, the threshold is set only according to the fluctuation range. If the object is heavy, it may be set according to the weight of the object. Alternatively, if the weight of the object is constant or the change width is small, a constant threshold value is used. Also good.

It is a block diagram of the robot hand apparatus which concerns on this Embodiment. It is a structural diagram of the robot hand of the robot hand device according to the present embodiment. It is a figure which shows a robot coordinate system and a hand coordinate system. It is explanatory drawing of the attitude | position of a robot hand, (a) is an example which shows a structure of the freedom degree of an arm, (b) is a link coordinate system in each joint of the arm shown in (a), a robot coordinate system, a hand It is a figure which shows the relationship of a coordinate system. It is a figure which shows the time change of the fingertip resultant force of a perpendicular direction. It is a figure which shows the relationship between the estimated value of an object weight, and a floor reaction force threshold value. It is a figure which shows the time change of the fingertip resultant force of the perpendicular direction when placing an object. It is a flowchart which shows operation | movement of the robot hand apparatus of FIG. It is a flowchart which shows the object weight estimation control in the control apparatus of FIG. It is a flowchart which shows the control which puts the object in the control apparatus of FIG.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Robot hand apparatus, 2 ... Arm, 3 ... Robot hand, 3a ... Thumb, 3b ... Index finger, 3c ... Middle finger, 3d ... Ring finger, 4 ... Camera, 5 ... Rotary encoder, 6 ... 6 axial force sensor, 7 ... Control Device, 20, 30 ... motor driver, 21, 31 ... motor, 32 ... interlocking joint

Claims (3)

When placing an object gripped by a robot hand, if the reaction force received from the place where the object is placed exceeds a threshold, control to separate the object from the robot hand,
A robot hand apparatus characterized in that the threshold value is a value proportional to the weight of an object. The robot hand apparatus according to claim 1 , wherein the reaction force is calculated from a difference between a weight of the object and a force for gripping the object in the robot hand. Converting the fingertip resultant force in the coordinate system of the robot hand into a fingertip resultant force in the base coordinate system of the robot hand device according to the posture of the robot hand from the coordinate system of the base of the robot hand device to the coordinate system of the robot hand; 3. The robot hand apparatus according to claim 1, wherein the weight of the object is estimated based on a fingertip resultant force in a vertical direction in a fingertip resultant force in a base coordinate system of the apparatus.

Images