Robot hand control device, robot hand, robot hand control method, and program

Abstract

Provided are a robot hand control device, a robot hand, a robot hand control method, and a program that are capable of performing force control whereby a gripped object is not dropped even if there is a sudden change in a gripping width of the gripped object. The robot hand control device is provided with a force control means that, when an object is being gripped by means of a hand mechanism having a plurality of fingers, changes the response speed of relative movement between the plurality of fingers when a gripping width, which is the interval between the plurality of fingers, increases or decreases.

Description

Robot hand control device, robot hand, robot hand control method, and program

[0001] The present disclosure relates to a robot hand control device, a robot hand, a robot hand control method, and a program.

[0002] Various types of robot hands, such as two-finger grippers, have been developed as end effectors for robot arms and are used in factory work, etc. If a robot hand has a manipulation function that allows it to change its posture while gripping an object in assembly work or grounding / display work, it is expected to be able to handle a wider range of tasks and improve work efficiency.

[0003] Regarding gripping by a robot hand, Patent Document 1 discloses a technology in which a parallel gripper structure with a parallel link mechanism controls the position of one finger to move the gripping position, and the other finger follows by force control, thereby rotating the posture of the gripped object.

[0004] Japanese Patent Application Laid-Open No. 2019-171532

[0005] However, with the technology described in Patent Document 1, when rotating a grasped object, if an attempt is made to rotate the posture of an object with a large aspect ratio between its long and short sides, such as a rectangular parallelepiped, the technology is unable to keep up with sudden changes in the grasping width, and there is a risk that the grasped object will fall.

[0006] The object of the present disclosure is to provide a robot hand control device, a robot hand, a robot hand control method, and a program that can perform force control that prevents a grasped object from dropping even if the grasped width of the grasped object changes suddenly.

[0007] The robot hand control device disclosed herein includes a force control means that, when grasping an object with a hand mechanism having multiple fingers, changes the response speed of the relative movement between the multiple fingers depending on whether the grasping width, which is the distance between the multiple fingers, is widening or narrowing.

[0008] The robot hand of the present disclosure comprises a hand mechanism having a plurality of fingers and the above-described robot hand control device.

[0009] The robot hand control method disclosed herein controls a hand mechanism having multiple fingers to grasp an object, and changes the response speed of the relative movement between the multiple fingers depending on whether the grasping width, which is the distance between the multiple fingers, is widening or narrowing.

[0010] The program disclosed herein causes a computer to execute the robot hand control method described above.

[0011] According to the present disclosure, when a grasped object is being held, even if there is a sudden change in the gripping width, force control can be performed to prevent the grasped object from dropping.

[0012] FIG. 1 is a schematic diagram showing an example of the configuration of a robot hand according to the first embodiment. FIG. 2 is a schematic diagram showing an example of the opening and closing operation of the robot hand of FIG. 1. FIG. 3 is a schematic diagram showing an example of the opening and closing operation of the robot hand of FIG. 1. FIG. 4 is a block diagram showing an example of the configuration of a robot hand control means according to the first embodiment. FIG. 5 is a schematic diagram showing the variable gain K of the variable gain compensator of FIG. 4. ｓ FIG. 6 is a graph for explaining the operation of the robot hand according to the first embodiment. FIG. 7 is a flowchart showing an example of a processing flow for the operation of the robot hand shown in FIG. 6. FIG. 8 is a block diagram showing an example of the configuration of a robot hand control means according to the second embodiment. FIG. 9 is a block diagram showing an example of the configuration of a robot hand control means according to the third embodiment. FIG. 10 is a schematic diagram for explaining the case of an operation in which a grasped object comes into contact with the external environment.

[0013] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. The present disclosure is not limited to the following embodiments, and various modifications are possible without departing from the spirit of the present disclosure. Furthermore, the present disclosure includes all possible combinations of the configurations shown in the following embodiments. In addition, in each drawing, components with the same reference numerals are the same or equivalent, and this is common throughout the entire specification.

[0014] Before describing the embodiments of the present invention in detail with reference to the drawings, various aspects of the present invention will be described below.

[0015] A robot hand control device according to one aspect of the present disclosure includes force control means that, when grasping an object with a hand mechanism having multiple fingers, changes the response speed of the relative movements between the multiple fingers depending on whether the grasping width, which is the distance between the multiple fingers, is widening or narrowing. According to this aspect, the robot hand can achieve stable force control by increasing the response speed for movements necessary for grasping, preventing the robot hand from losing force and dropping the grasped object, and by excessively increasing the response speed for movements unnecessary for grasping, preventing self-oscillation.

[0016] In the above aspect, the force control means may change the response speed of the finger movement by changing a feedback gain for the grip force error depending on whether the grip force error, which is the difference between the target grip force and the grip force, is positive or negative. According to this aspect, it is possible to easily change the response speed of the finger movement in the direction of widening and narrowing the grip width.

[0017] In the above aspect, the force control means may increase the feedback gain as the absolute value of the grip force error increases in the direction narrowing the grip width. According to this aspect, by increasing the feedback gain for the grip force error as the absolute value of the grip force error increases, it is possible to improve responsiveness in a state where the grip force error is large and greater responsiveness is required.

[0018] In the above aspect, the force control means may increase the slope of the feedback gain with respect to the grip force error as the absolute value of the grip force error increases in the direction narrowing the grip width. According to this aspect, by increasing the slope of the feedback gain with respect to the grip force error, it is possible to more accurately increase responsiveness in a state where the grip force error is large and increased responsiveness is required.

[0019] In the above aspect, the force control means may maintain a constant feedback gain for the grip force error in the direction of widening the grip width. By maintaining a constant feedback gain, it is possible to prevent the robot hand from excessively widening the grip width, weakening its grip, and becoming unstable, which may result in self-excited oscillation or dropping the gripped object.

[0020] In the above aspect, the force control means may change the feedback gain so that the feedback gain in the direction narrowing the grip width and the feedback gain in the direction widening the grip width match when the grip force error is 0. According to this aspect, the force control does not become discontinuous when switching between the direction narrowing the grip width and the direction widening the grip width, and stable force control is possible.

[0021] <First Embodiment> A first embodiment of the present disclosure will now be described. Fig. 1 is a schematic diagram showing an example of the configuration of a robot hand 1 according to the first embodiment. The robot hand 1 is a two-fingered gripper, and includes a hand base 2, finger mechanisms 3R and 3L including parallel link mechanisms 6R and 6L and fingertip mechanisms 7R and 7L, and a robot hand control means 8 (see Fig. 4).

[0022] The hand base 2 is formed with a wrist connection section 5, which is connected to a multi-degree-of-freedom robot arm or the like. This allows the robot hand 1 to function as the end effector of the robot arm. Finger drive servomotors 9R and 9L are also disposed on the hand base 2, and parallel link mechanisms 6R and 6L are connected to their output shafts. The hand base 2 also includes finger joints 4R and 4L.

[0023] The parallel link mechanisms 6R and 6L include active links 10R and 10L that are connected to the output shafts of the finger drive servomotors 9R and 9L and transmit driving force, and passive links 11R and 11L that are not connected to the output shafts of the finger drive servomotors 9R and 9L.

[0024] Fingertip mechanisms 7R and 7L are connected to the ends of the parallel link mechanisms 6R and 6L. The fingertip mechanisms 7R and 7L include finger skeletons 12R and 12L, finger inner translational drive mechanisms 13R and 13L, and force sensors 14Rf, 14Rb, 14Lf, and 14Lb.

[0025] The inner finger translational drive mechanisms 13R and 13L drive translational drive belts 15R and 15L that pass through the insides of the finger skeletons 12R and 12L. Both ends of the translational drive belts 15R and 15L are fixed to motor-side take-up rollers 17R and 17L and spring-side take-up rollers 19R and 19L, respectively. The translational drive belts 15R and 15L are also routed so as to wrap around passive rollers 20R and 20L disposed at the tips of the finger skeletons 12R and 12L. The motor-side take-up rollers 17R and 17L are connected to the output shafts of translational drive mechanism servomotors 16R and 16L disposed on the finger skeletons 12R and 12L. The spring-side take-up rollers 19R and 19L are connected to the output shafts of translational drive mechanism helical springs 18R and 18L.

[0026] Between the finger skeletons 12R and 12L and the translation drive belts 15R and 15L, there are arranged a pressure-sensitive sheet (not shown) that outputs a voltage according to pressure, and force sensors 14Rf, 14Rb, 14Lf, and 14Lb that are composed of voltage circuits.

[0027] 2 and 3 are schematic diagrams showing an example of the opening and closing operation of the robot hand 1 in FIG. 1. FIG. 2 shows the robot hand 1 in an open state, and FIG. 3 shows the robot hand 1 in a closed state. With the robot hand 1 having the above-described mechanical configuration, when the finger drive servomotors 9R and 9L are driven, the parallel link mechanisms 6R and 6L swing, and the fingertip mechanisms 7R and 7L move in parallel. This allows the opening and closing operation shown in FIGS. 2 and 3 to be realized.

[0028] Furthermore, by driving the translational drive mechanism servo motors 16R and 16L, the translational drive belts 15R and 15L are wound onto the motor-side take-up rollers 17R and 17L, and move translationally from the fingertip side to the back side. On the other hand, by rotating the translational drive mechanism servo motors 16R and 16L in the reverse direction to the above, the translational drive belts 15R and 15L are released from the motor-side take-up rollers 17R and 17L, and are wound onto the spring-side take-up rollers 19R and 19L due to the rotational tension of the translational drive mechanism winding springs 18R and 18L, and move translationally from the back side to the fingertip side.

[0029] 4 is a block diagram showing an example of the configuration of the robot hand control means 8 according to the first embodiment. As shown in FIG. 4, the robot hand control means 8 includes a control computer 35, which performs various processes in each block shown in FIG. 4 at a fixed control period, such as 1 msec. The robot hand control means 8 receives a control command τ ｃ By outputting the above, the position h and gripping force f of the fingertip mechanisms 7R and 7L, and the position b of the translational drive belts 15R and 15L are controlled.

[0030] Here, the control command τ ｃ is "τ ｃ =(τ ｈＲｄ , τ ｈＬｄ , τ ｂＲｄ , τ ｂＬｄ The position h of the fingertip mechanisms 7R and 7L is expressed as "h = (h Ｒ , h Ｌ ) and the position b of the translational drive belts 15R and 15L is "b = (b Ｒ , b Ｌ )"

[0031] The motion control means 21 is composed of a motion sequence control means 38 and a target trajectory generation means 39, and is used to determine the target positions h of the fingertip mechanisms 7R and 7L. ｄ and the force control target force f ｄ and the target position b of the translation drive belts 15R and 15L. ｄ Here, the target position h of the fingertip mechanisms 7R and 7L is output. ｄ is "h ｄ= (h Ｒｄ , h Ｌｄ The target position bd of the translational drive belts 15R and 15L is "b ｄ = (b Ｒｄ , b Ｌｄ )"

[0032] The operation sequence control means 38 receives an operation instruction C for the robot hand 1 from a host system 41 such as a robot arm control system. ｇ through the communication interface, executes the operation sequence of the robot hand 1, and sends the operation result as an operation notification R ｈ The operation sequence control means 38 notifies the upper system 41 of the target arrival time t ｐｈ , target arrival position h ｐｄ , the target arrival time t of the translation drive belts 15R and 15L ｐｂ , target arrival position b ｐｄ to the target trajectory generating means 39. Furthermore, the operation sequence control means 38 outputs the force control target force f ｄ is output to the force control means 40.

[0033] As a result, the operation sequence control means 38 performs the movement operation to the home position after startup, and the operation instruction C from the higher-level system 41. ｇ The opening and closing operations of the fingertip mechanisms 7R and 7L and the operation of the translation drive belts 15R and 15L are controlled based on the above. Here, the target position h of the fingertip mechanisms 7R and 7L is ｐｄ is "h ｐｄ = (h Ｒｐ , h Ｌｐ The target position b of the translation drive belts 15R and 15L is ｐｄ is "b ｐｄ = (b Ｒｐ , b Ｌｐ )"

[0034] The target trajectory generating means 39 controls the position h of the fingertip mechanisms 7R and 7L to the target position h ｐｄ Target position h for operation to ｄ , and the position b of the translation drive belts 15R and 15L is set to the target position b ｐｄ Target belt position b for operation to ｄis generated every moment as time series data using polynomial interpolation.

[0035] Here, the position h of the fingertip mechanisms 7R and 7L is expressed as "h = (h R , h L ) and the target positions h ｄ is "h ｄ = (h Rｄ , h Lｄ The position b of the translational drive belts 15R and 15L is expressed as "b = (b R ,b L ) and the target position b ｄ is "b ｄ = (b Ｒｄ , b Ｌｄ )

[0036] The finger inverse kinematics calculation means 22 calculates the target positions h of the fingertip mechanisms 7R and 7L. ｄ The target finger joint angle q is calculated by inverse kinematics calculation. ｈｄ Convert to the target finger joint angle q ｈｄ is "q ｈｄ = (q ｈＲｄ , q ｈＬｄ )"

[0037] The finger joint angle error calculation means 23 calculates the finger joint angle error q ｈｅ Calculate the finger joint angle error q ｈｅ is "q ｈｅ = (q ｈＲｅ , q ｈＬｅ ) and the target finger joint angle q ｈｄ From the current finger joint angle q ｈ That is, the finger joint angle error q ｈｅ is "q ｈｅ =q ｈｄ -q ｈ ". The current finger joint angle q ｈ is output from the robot hand I / O 34, which will be described later, and is ｈ = (q ｈＲ , q ｈＬ )"

[0038] The finger joint angle error compensation means 24 is a so-called PD controller, and calculates a finger joint angle control command torque τ ｈ In equation (1), "K ｐ " is the proportional gain, and "K ｄ " is the differential gain. τ ｈ =K ｐ q ｈｅ +K ｄ dq ｈｅ / dt ... (1)

[0039] The force sensors 14Rf, 14Rb, 14Lf, and 14Lb output a voltage V according to the applied force. ｆＡ is outputted, and is converted into a digital value V by the A / D converter 37. ｆＤ The digital value of the voltage V ｆＤ is the force f ｓ and input to the force control means 40 as the gripping force f.

[0040] Here, the voltage V ｆＡ is "V ｆＡ = (V ｆＡＲ , V ｆＡＬ ) and the voltage V ｆＡ The digital value V ｆＤ is "V ｆＤ = (V ｆＤＲ , V ｆＤＬ )) Also, the force f ｓ is "f ｓ = (f Ｒ , f Ｌ ) and the gripping force f is expressed as "f = (f Ｒ +f Ｌ ) / 2".

[0041] The force control means 40 is composed of a gripping force error calculation means 42 , a gripping force compensator 25 , a variable gain compensator 26 , and a force-torque conversion means 27 .

[0042] The gripping force error calculation means 42 calculates the gripping force error f. ｅ Calculate the gripping force error f ｅ is the force control target force f ｄ The gripping force error f is subtracted from the ｅis "f ｅ = f ｄ -f".

[0043] The gripping force compensator 25 calculates the gripping force compensation output f based on the equation (2). ＰＤ In equation (2), "K ｐｆ " is the proportional gain, and "K ｄｆ " is the differential gain. f ＰＤ =K ｐｆ f ｅ +K ｄｆ df ｅ / dt ... (2)

[0044] The variable gain compensator 26 calculates the gripping force variable compensation output f based on the equation (3). Ｋｓ Calculate f Ｋｓ =K s f eＰＤ ...(3)

[0045] In formula (3), "K s " is a variable gain, which changes according to the characteristics shown in the graph of FIG. 5, which will be described later. In FIG. 5, the gripping force error f ｅ "f ｅ In the "<0" region, the variable gain K ｓ is "K ｓ On the other hand, the gripping force error f ｅ "f ｅ In the ">=0" region, the gripping force error f ｅ The larger the variable gain K ｓ For example, "K" becomes an exponentially large value. ｓｍａｘ " is variable gain K ｓ The maximum value of the variable gain K ｓ corresponds to the "feedback gain" in this disclosure.

[0046] The force-torque conversion means 27 outputs a variable gripping force compensation output f Ｋｓ is input, and the input gripping force variable compensation output f Ｋｓ Based on this, the control torque τ ｆ First, the force-torque conversion means 27 converts the input variable gripping force compensation output f Ｋｓ Based on this, (f Ｒｃ , f Ｌｃ ) to (f Ｋｓ, -f Ｋｓ ) into (f Ｒｃ , f Ｌｃ ) are the force control commands for the right finger mechanism 3R and the left finger mechanism 3L, respectively.

[0047] Next, the force-torque converting means 27 converts the torque (f Ｋｓ , -f Ｋｓ ) to obtain the control torque τ ｆ In equation (4), "l ｐ " is the link length of the parallel link mechanisms 6R and 6L. τ ｆ = (f Ｒｃ l ｐ / cosq ｈｒ , f Ｌｃ l ｐ / cosq ｈＬ ) ... (4)

[0048] The position / force control selection means 28 switches the output value based on the position / force control selection command s from the motion control means 21. When the position / force control selection command s is 0 (s=0), the position / force control selection means 28 selects the finger joint angle control command torque τ ｈ In this way, position control for controlling the finger position is performed. When the position / force control selection command s is 1 (s=1), the position / force control selection means 28 outputs the control torque τ ｆ As a result, when the position / force control selection command s is 1, force control for controlling the gripping force is performed.

[0049] FIG. 5 shows the variable gain K of the variable gain compensator 26 in FIG. ｓ 10 is a graph for explaining the gripping force error f ｅ The vertical axis represents the variable gain K ｓ Shows.

[0050] Gripping force error f ｅ When is negative (f ｅ <0), variable gain K ｓ is a constant value of "1". On the other hand, the gripping force error f ｅ If is positive (f ｅ ≧0), variable gain K ｓ is the gripping force error is 0 (f ｅ= 0), and the gripping force error f ｅ The larger it gets, the larger it becomes exponentially.

[0051] According to the characteristics shown in FIG. 5, the gripping force error f ｅ If is negative, the gripping force f is equal to the target gripping force f ｄ Since the value is larger, the fingers open and the grip width is increased. ｓ = 1, so the force control command value f ｃ is the output of the gripping force compensator 25, and is subjected to force control by the gripping force compensator 25.

[0052] On the other hand, gripping force error f ｅ If is positive, the gripping force f is equal to the target gripping force f ｄ Since the value is smaller than the above, the fingers close together to reduce the grip width. Furthermore, since the variable gain Ks has an exponential characteristic, the grip force error f ｅ The larger the absolute value of f, the larger the variable gain Ks becomes, and the force control command value f ｃ Since the value of the force is large, the force control has a higher gain and a higher response than the PD compensator.

[0053] In this example, the gripping force error f ｅ The variable gain K when is positive ｓ Although the characteristics of the gripping force error f are described as being exponential, this is not limited to this example. ｅ The variable gain K when is positive ｓ The characteristics of the gripping force error f ｅ The larger the value of the variable gain K ｓ Furthermore, for example, the gripping force error f ｅ The variable gain K when is positive ｓ The characteristic of the gripping force error f ｅ Regardless of the magnitude of ｓ Alternatively, the characteristic may be such that the rate of increase is constant.

[0054] The belt inverse kinematics calculation means 29 in FIG. 4 calculates the target belt position b ｄ The target winding roller angle q ｂｄConvert to target belt position b ｄ is "b ｄ = (b Ｒｄ , b Ｌｄ )) Also, the target winding roller angle q ｂｄ is "q ｂｄ = (q ｂＲｄ , q ｂＬｄ )"

[0055] The belt roller angle error calculation means 30 calculates the belt roller angle error q ｂｅ Calculate the belt roller angle error q ｂｅ is "q ｂｅ = (q ｂＲｅ , q ｂＬｅ ) and the standard winding roller angle q ｂｄ From the current belt roller angle q ｂ That is, the belt roller angle error q ｂｅ is "q ｂｅ =q ｂｄ -q ｂ ". The current belt roller angle q ｂ is output from the robot hand I / O 34, which will be described later, and is ｂ = (q ｂＲ , q ｂＬ )"

[0056] The belt roller angle error compensation means 31 is a so-called PD controller, and calculates the belt roller angle control command force τ ｂ In equation (5), "K ｂｐ " is the proportional gain, and "K ｂｖ " is the differential gain. τ ｂ =K ｂｐ q ｂｅ +K ｂｄ dq ｂｅ / dt ... (5)

[0057] The above control command τ ｈｄ or τ ｆ , and τ ｂ is input to the robot hand I / O 34, and is sent to the motor drivers 36hR, 36hL, 36bR, and 36bL as a control command τ ｃ =(τ ｈＲｄ , τｈＬｄ , τ ｂＲｄ , τ ｂＬｄ ) This drives the finger drive servomotors 9R and 9L and the translation drive mechanism servomotors 16R and 16L.

[0058] The robot hand I / O 34 also receives input of encoder values ​​of the finger drive servo motors 9R and 9L and the translation drive mechanism servo motors 16R and 16L transmitted from the motor drivers 36hR, 36hL, 36bR, and 36bL. The robot hand I / O 34 then converts the input encoder values ​​into the current finger joint angle q ｈ = (q ｈＲ , q ｈＬ ), current belt roller angle q ｂ = (q ｂＲ , q ｂＬ ) and output.

[0059] The finger sequence kinematics calculation means 32 calculates the current finger joint angle q ｈ is converted to the current finger position h. The current finger joint angle q ｈ is "q ｈ = (q ｈＲ , q ｈＬ )) and the current finger joint angle h is "h = (h Ｒ , h Ｌ )"

[0060] The belt order kinematics calculation means 33 calculates the current belt roller angle q ｂ Converts the current belt position b into the current belt roller angle q ｂ is "q ｂ = (q ｂＲ , q ｂＬ The current belt position b is expressed as "b = (b Ｒ , b Ｌ )"

[0061] According to the feedback control system shown in FIG. 4 described above, when the position / force control selection command s is 0, the current finger position h=(h Ｒ , h Ｌ ) is the target position h ｄ = (h Rｄ , h Lｄ) When the position / force control selection command s is 1, the current gripping force f by the fingertip mechanisms 7R and 7L is controlled to be equal to the target gripping force f ｄ Furthermore, the current belt position b of the translation drive belts 15R and 15L is controlled as b=(b Ｒ , b Ｌ ) is the target belt position b ｄ = (b Ｒｄ , b Ｌｄ ) is controlled.

[0062] The robot hand control means 8 is configured in hardware terms with a control computer 35 that executes a control program, motor drivers 36hR, 36hL, 36bR, and 36bL, and an A / D converter 37. The parts of the hand control means 8 other than the motor drivers 36hR, 36hL, 36bR, and 36bL and the A / D converter 37 are realized in software terms as a control program executed by the control computer 35. Therefore, for example, each step described below can be executed by storing a computer program having steps that constitute each operation in a readable manner on a recording medium such as a storage device, such as a hard disk, and loading the computer program into a temporary storage device, such as a semiconductor memory, of the computer and executing it using a CPU.

[0063] Such robot hand control means 8 may be disposed in the robot hand 1 itself, or may be disposed in an external information terminal, etc. When the robot hand control means 8 is disposed in an external information terminal, etc., the robot hand 1 can be remotely controlled by communicating between the information terminal and the robot hand 1 using a communication means (not shown).

[0064] Angle pulse information p output from the encoders of the finger drive servomotors 9R and 9L, which are angle sensors for the finger joints 4R and 4L, the motor side winding rollers 17R and 17L, and the translation drive mechanism servomotors 16R and 16L. ｈＲ , p ｈＬ , p ｂＲ and p ｂＬis input to the motor drivers 36hR, 36hL, 36bR and 36bL, and the angle information α ｈＲ , α ｈＬ , α ｂＲ and α ｂＬ and output to the robot hand control means 8.

[0065] The robot hand control means 8 calculates control command values ​​for the finger joints 4R and 4L and the motor-side take-up rollers 17R and 17L based on the acquired angle information. The calculated control command values ​​are provided to motor drivers 36hR, 36hL, 36bR, and 36bL. These motor drivers 36hR, 36hL, 36bR, and 36bL drive the finger drive servomotors 9R and 9L and the translation drive mechanism servomotors 16R and 16L of the robot hand 1.

[0066] Motor drivers 36hR, 36hL, 36bR and 36b Ｌ , and the A / D converter 37 are connected to the control computer 35 via a network communication such as EtherCAT (registered trademark), allowing the exchange of angle information and control command values.

[0067] The operation of the robot hand 1 will be described with reference to Fig. 6 and Fig. 7. Fig. 6 is a schematic diagram for explaining the operation of the robot hand 1 according to the first embodiment. Fig. 7 is a flowchart showing an example of the processing flow for the operation of the robot hand 1 shown in Fig. 6.

[0068] In step S1, after the robot hand 1 is started, the origin return function of the motor drivers 36hR, 36hL, 36bR, and 36bL operates the fingertip mechanisms 7R and 7L and the translation drive belts 15R and 15L to their movable mechanism limits and then stops them (operation of S-1 in FIG. 6).

[0069] In step S2, the robot hand I / O 34 receives the state where the robot hand 1 has stopped at the limit of its movable mechanism, and calculates the current finger joint angle q so that the center line of the robot hand 1 coincides with the origin of the fingertip mechanisms 7R and 7L, and the center of the entire length of the translational drive belts 15R and 15L coincides with the origin. ｈ = (qｈＲ , q ｈＬ ), and belt roller angle q ｂ = (q ｂＲ , q ｂＬ ) and output.

[0070] In step S3, after the movement to the movable mechanism limit is completed, the operation control means 21 moves the fingertip mechanisms 7R and 7L to the home position h Ｈ , and the translation drive belts 15R and 15L are moved to the home position b Ｈ An action to move to is generated and executed (the action of S-3 in FIG. 6).

[0071] In addition, the operation sequence control means 38 ｈ The initial value of R ｈ = 0 to R ｈ = 1, and notify the host system 41 that the preparation of the robot hand 1 is complete. Ｈ is "h Ｈ = (h ＨＲ , h ＨＬ In addition, the home position b of the translation drive belts 15R and 15L is Ｈ is "b Ｈ = (0, 0)".

[0072] Then, in step S4, the robot hand 1 is moved to a position where the robot arm can grasp the object, and the operation command from the host system 41 is set to the initial value C ｇ = 0 to C ｇ When the value is switched to 1, the process proceeds to step S5 (operation of S-4 in FIG. 6).

[0073] In step S5, the operation sequence control means 38 outputs s=1 as a position / force control selection command to switch the control system 41 to grip force control.

[0074] In step S6, the position / force control selection means 28 selects the control torque τ ｆ Select the torque command τ ｄ The fingertip mechanisms 7R and 7L are operated by outputting the force f ｄ (operation S-6 in FIG. 6).

[0075] In step S7, the operation sequence control means 38 determines the arrival time t ｂｐｕ , target arrival position b ｐ = (b ｐｕ , b ｐｕ ) to operate the translation drive belts 15R and 15L, and the grasped object is moved by a distance b ｐｕ (operation S-7 in FIG. 6).

[0076] In step S8, the operation sequence control means 38 determines the arrival time t ｂｐｒ , target arrival position b ｐ = (b ｐｕ +b ｐｒ , b ｐｕ -b ｐｒ ) and the translation drive belts 15R and 15L are driven by b ｐｒ By performing the operation in the opposite direction at each time, the posture of the grasped object is rotated (operation S-8-1 in FIG. 6).

[0077] If the aspect ratio of the object, which is the ratio between its sides, is large, the grip width decreases rapidly as the object rotates (operation S-8-2 in Figure 6). Furthermore, if the fingers cannot keep up with the sudden decrease in grip width, the grip force decreases, the friction between the belt and the object is lost, and the object may fall. To address this issue, the force control unit 40 utilizes the variable gain compensator 26, shown by the gain characteristics in Figure 5, which responds quickly to the decrease in grip force and operates in a direction that closes the fingers and rapidly narrows the grip width, thereby achieving rotation of the object's orientation without dropping it.

[0078] In step S9, when the posture change of the grasped object is completed, the operation sequence control means 38 sends a posture change completion signal R to the host system 41. ｈ = 2. Then, the host system 41 moves the tip end position of the robot arm to the target position where the object to be grasped is to be placed (operation of S-9 in FIG. 6).

[0079] Then, in step S10, the host system 41 sends a grip release signal C ｇ When the grip release signal is received, the process proceeds to step S11.

[0080] In step S11, the operation sequence control means 38 ｐｂ = (b Ｒｓ -b ｐｕ , b Ｌｓ -b ｐｕ ), t ｐｂ = t ｏｐ and the translation drive belts 15R and 15L are operated to move the grasped object t ｏｐ The fingertip is moved to the touch panel in the movement time (operation S-11 in FIG. 6).

[0081] In step S12, the operation sequence control means 38 outputs s=0 to switch the control of the fingertip mechanisms 7R and 7L to position control, and ｐｂ = (h ＨＲ , h ＨＬ ), t ｐｈ = t Ｈ As a result, the fingertip mechanisms 7R and 7L output t Ｈ The robot moves to the home position in the movement time and releases the grasped object (operation of S-12 in FIG. 6).

[0082] In step S13, the operation sequence control means 38 sends a grip release completion signal R Ｈ = 3. Then, the process returns to step S4 and proceeds to the next operation.

[0083] As described above, in the first embodiment, the force control means 40 has the variable gain compensator 26 whose gain characteristic is shown in Fig. 5, thereby changing the response speed of the finger movement in the direction of widening and narrowing the grip width. In particular, in this example, by speeding up the finger movement in the direction of narrowing the grip width, the fingers respond quickly to a decrease in grip force, closing and moving in the direction of rapidly narrowing the grip width.

[0084] Furthermore, as shown in the gain characteristics of Figure 5, rather than simply increasing the overall gain, the operation in the direction of widening the gripping width is performed with the gain of the normal PD compensator 25, and in the direction of narrowing the gripping width, the greater the decrease in gripping force, the higher the gain is set, thereby suppressing the occurrence of self-oscillation while enabling high-speed response where necessary.

[0085] Due to these effects, when the finger inner translation drive mechanisms 13R and 13L rotate the posture of the grasped object, the grasped object can be rotated in a stable posture without dropping even if a sudden change in the gripping width occurs.

[0086] The robot hand 1 according to the first embodiment is not limited to the above example, and may, for example, speed up the finger movement in the direction of widening the grip width. Specifically, when gripping both ends of the short side of an object such as a rectangular parallelepiped with a large aspect ratio and rotating the posture, the fingers can be made to follow a sudden change in the grip width by speeding up the finger movement in the direction of widening the grip width. In this case, the grip force error f ｅ When is negative (f ｅ <0) variable gain K ｓ , the gripping force error f ｅ The larger the absolute value of is, the larger the value is set.

[0087] In this way, with the robot hand 1 according to the first embodiment, the force control means 40 changes the response speed of the finger movements in the direction of widening and narrowing the grip width, thereby configuring a control system that increases the feedback gain of force control only for necessary movements, thereby enabling the fingers to follow sudden changes in the grip width.

[0088] Second Embodiment Hereinafter, a second embodiment of the present disclosure will be described. In the following description, parts common to the first embodiment will be denoted by the same reference numerals, and detailed description thereof will be omitted.

[0089] 8 is a block diagram showing an example of the configuration of the robot hand control means 8 according to the second embodiment. The robot hand control means 8 according to the second embodiment further includes a finger center position error compensator 44, and calculates the finger center position h in addition to the grip force f. ｃ Position control is also possible at the same time.

[0090] The target trajectory generating means 39 outputs the target finger center position value h ｃｄ is output, and the finger sequence kinematics calculation means 32 calculates the finger center position h in addition to the finger position h. ｃ = (h Ｒ +h Ｌ) / 2 is output and input to the finger center position error calculation means 43.

[0091] The finger center position error calculation means 43 calculates the finger center position error h ｃｅ Calculate the finger center position error h ｃｅ is the target finger center position h ｃｄ From the center of the fingers ｃ That is, the finger center position error h ｃｅ is "h ｃｅ =h ｃｄ -h ｃ "

[0092] The finger center position error compensator 44 is a so-called PD controller, and generates a finger center position compensation output f based on the equation (6). ｈｃ In equation (6), "K ｐｈｃ " is the proportional gain, and "K ｄｈｃ " is the differential gain. f ｈｃ =K ｐｈｃ h ｃｅ +K ｄｈｃ dh ｃｅ / dt ... (6)

[0093] The finger center position compensation output f calculated by the finger center position error compensator 44 ｈｃ and the gripping force compensation output f calculated by the variable gain compensator 26. Ｋｓ is input to the force-torque conversion means 27.

[0094] The force-torque conversion means 27 converts the input finger center position compensation output f ｈｃ and gripping force compensation output f Ｋｓ Based on this, the control torque τ ｆ The force-torque conversion means 27 converts the input finger center position compensation output f ｈｃ and gripping force compensation output f Ｋｓ Based on this, (f Ｒｃ , f Ｌｃ ) to (f ｈｃ +f Ｋｓ , f ｈｃ -f Ｋｓ ) and convert it into the torque command τ ｆ This allows for force control.

[0095] The finger-belt interlocking means 45 interlocks the movement of the fingers with the movement of the belt. When the middle positions of the left and right fingers are displaced from the center position of the hand, the positions of the left and right fingertips will be displaced in the length direction of the fingers due to the parallel link mechanism.

[0096] To cope with this, the finger-belt interlocking means 44 adjusts the positions of the left and right belts based on the equation (7) as follows: ｂｄｍ In equation (7), the force is adjusted to "(q ｈＲｃ , q ｈＬｃ ) is the finger joint angle (q ｈＲ , q ｈＬ ) Also, "l ｐ " is the link length of the parallel link mechanisms 6R and 6L. ｂｄｍ = (q ｂｄＲ +l ｐ cosq ｈＲ -l ｐ cosq ｈＲｃ , q ｂｄＬ +l ｐ cosq ｈＬ -l ｐ cosq ｈＬｃ ) ... (7)

[0097] As described above, the robot hand 1 according to embodiment 2 can respond quickly to a decrease in gripping force and can rotate the gripped object in a stable posture without dropping it, similar to embodiment 1. Furthermore, by including the finger center position error compensator 44, the robot hand 1 according to embodiment 2 can simultaneously control the gripping force and the finger center position.

[0098] Third Embodiment Hereinafter, a third embodiment of the present disclosure will be described. In the following description, parts common to the first and second embodiments will be denoted by the same reference numerals, and detailed description thereof will be omitted.

[0099] 9 is a block diagram showing an example of the configuration of the robot hand control means 8 according to the present embodiment 3. The robot hand control means 8 according to the present embodiment 3 has an external force compensator 48, and is capable of simultaneously controlling the external force acting on the object to be grasped in addition to the gripping force f.

[0100] In the third embodiment, the operation sequence control means 38 outputs the external force target value f ｅｘｄ and the gripping force target value f ｄ is output.

[0101] The gripping force / external force calculation means 46 receives the force f from the robot hand I / O 34. ｓ = (f Ｒ , f Ｌ ) is input. The gripping force / external force calculation means 46 calculates the external force f based on the formulas (8) and (9). ｅｘ and gripping force f are calculated respectively. ｅｘ = (f Ｒ -f Ｌ ) / 2...(8) f=(f Ｒ +f Ｌ ) / 2 ... (9)

[0102] The gripping force error calculation means 47 calculates the gripping force error f ｅｘｅ Calculate the gripping force error f ｅｘｅ is the external force target value f ｅｘｄ From external force f ｅｘ That is, the gripping force error f ｅｘｅ is "f ｅｘｅ = f ｅｘｄ -f ｅｘ The calculated gripping force error f ｅｘｅ is input to the external force compensator 48.

[0103] The external force compensator 48 is a so-called PD controller, and generates an external force compensation output f based on the equation (10). ｅｘｃ In equation (10), "K ｐｅｘ " is the proportional gain, and "K ｄｅｘ " is the differential gain. f ｅｘｃ =K ｐｅｘ f ｅｘｅ +K ｄｅｘ df ｅｘｅ / dt...(10)

[0104] The external force compensation output f calculated by the external force compensator 48 ｅｘｃ and the gripping force compensation output f calculated by the variable gain compensator 26. Ｋｓ is input to the force-torque conversion means 27.

[0105] The force-torque conversion means 27 converts the input external force compensation output f ｅｘｃ and gripping force compensation output f Ｋｓ Based on this, the control torque τ ｆ The force-torque conversion means 27 converts the input external force compensation output f ｅｘｃ and gripping force compensation output f Ｋｓ Based on this, (f Ｒｃ , f Ｌｃ ) to (f ｅｘｃ +f Ｋｓ , f ｅｘｃ -f Ｋｓ ) and convert it into the torque command τ ｆ This allows for force control.

[0106] In this way, in the third embodiment, in the case of an operation in which a grasped object comes into contact with the external environment as shown in FIG. 10, the grasping force f is controlled, and the external force f ｅｘ This makes it possible to control the following.

[0107] As described above, the robot hand 1 according to the third embodiment can respond quickly to a decrease in gripping force and can rotate the gripped object in a stable posture without dropping it, similar to the robot hand 1 according to the first and second embodiments. Furthermore, by including the external force compensator 48, the robot hand 1 according to the third embodiment can simultaneously control the gripping force and the external force.

[0108] Although the first to third embodiments have been described above, the present disclosure is not limited to the first to third embodiments, and various modifications and applications are possible without departing from the spirit and scope of the present disclosure. In the first to third embodiments, the finger mechanisms 3R and 3L are described as being parallel link mechanisms 6R and 6L, but this is not limiting. For example, the finger mechanisms 3R and 3L may be other mechanisms, such as a linear motion mechanism that performs parallel finger drive. Even in this case, the same effects as those of the first to third embodiments can be achieved.

[0109] Furthermore, in the first to third embodiments, a case has been described in which a gripped object is rotated using a translational drive mechanism that uses the translational drive belts 15R and 15L, but this is not limited to this example. For example, control that changes the response speed of finger movement in response to changes in grip width can also be applied to a robot hand that is not provided with a translational drive mechanism. This allows for a high-speed response to a decrease in grip force, even when the grip width changes while gripping an object, and allows for stable rotation of the gripped object without dropping it.

[0110] The disclosures of the specification, drawings and abstract contained in Japanese Patent Application No. 2024-109518, filed on July 8, 2024, are incorporated herein by reference in their entirety.

[0111] The robot hand of the present disclosure is useful as a gripper hand that can be used to grasp an object as an end effector of a robot arm.

[0112] REFERENCE SIGNS LIST 1 Robot hand 2 Hand base 3R, 3L Finger mechanism 4R, 4L Finger joint 5 Wrist connection 6R, 6L Parallel link mechanism 7R, 7L Fingertip mechanism 8 Robot hand control means 9R, 9L Finger drive servomotor 10R, 10L Active link 11R, 11L Passive link 12R, 12L Finger skeleton 13R, 13L Finger inner translation drive mechanism 14Rf, 14Rb, 14Lf, 14Lb Force sensor 15R, 15L Translation drive belt 16R, 16L Translation drive mechanism servomotor 17R, 17L Motor side winding roller 18R, 18L Translation drive mechanism winding spring 19R, 19L Spring side winding roller 20R, 20L Passive roller 21 Operation control means 22 Finger inverse kinematics calculation means 23 Finger joint angle error calculation means 24 Finger joint angle error compensation means 25 Gripping force compensator 26 Variable gain compensator 27 Force-torque conversion means 28 Position / force control selection means 29 Belt inverse kinematics calculation means 30 Belt roller angle error calculation means 31 Belt roller angle error compensation means 32 Finger sequence kinematics calculation means 33 Belt sequence kinematics calculation means 34 Robot hand I / O 35 Control computer 36hR, 36hL, 36bR, 36bL Motor driver 37 A / D converter 38 Operation sequence control means 39 Target trajectory generation means 40 Force control means 41 Upper system 42 Gripping force error calculation means 43 Finger center position error calculation means 44 Finger center position error compensator 45 Finger-belt linkage means 46 Gripping force / external force calculation means 47 Gripping force error calculation means 48 External force compensator

Claims

1. A robot hand control device that is equipped with a force control means that, when grasping an object with a hand mechanism having multiple fingers, changes the response speed of the relative movement between the multiple fingers when the grasping width, which is the distance between the multiple fingers, increases or decreases.

2. A robot hand control device according to claim 1, wherein the force control means determines whether the gripping width should be increased or decreased based on the gripping force when gripping the object.

3. A robot hand control device as described in claim 2, wherein the force control means changes the response speed by changing a feedback gain for the gripping force error depending on whether the gripping force error, which is the difference between the target value of the gripping force and the gripping force, is positive or negative.

4. A robot hand control device according to claim 3, wherein the force control means increases the feedback gain as the absolute value of the grip force error increases when the grip width is narrowed.

5. A robot hand control device according to claim 4, wherein the force control means increases the rate of increase of the feedback gain as the absolute value of the grip force error increases when the grip width is narrowed.

6. A robot hand control device according to claim 3, wherein the force control means keeps the feedback gain constant when the gripping width is widened.

7. A robot hand control device as described in claim 3, wherein the force control means is configured such that when the grip force error is 0, the feedback gain when narrowing the grip width coincides with the feedback gain when widening the grip width.

8. A robot hand comprising: a hand mechanism having a plurality of fingers; and the robot hand control device according to claim 1.

9. A robot hand control method that controls a hand mechanism having multiple fingers to grasp an object, and changes the response speed of the relative movement between the multiple fingers when the grasping width, which is the distance between the multiple fingers, increases or decreases.

10. A program for causing a computer to execute the robot hand control method according to claim 9.

Images