End effector control method

Abstract

To provide an end effector control method which can smoothly achieve operation of an object with large temporal variations in work requiring large operation force.SOLUTION: An end effector control method is an end effector control method capable of operating an object, and includes the steps of: controlling the operation by a hybrid of a joint angle of an end effector and joint torque, acquiring a contact point position of the end effector and the target object by a sensor, and estimating force applied from the end effector to the target object, on the basis of the acquired contact position; estimating an error with an object operation model when operating the target object, from a position change of the force acting on the target object and the target object; and re-determining force required for the operation from an error amount estimated from the error with the object operation model.SELECTED DRAWING: Figure 10

Description

Technical Field

[0001] The present invention relates to an end effector control method.

Background Art

[0002] When controlling an end effector to perform work, an operation trajectory is determined in advance and the end effector operates according to it. When the error from the commanded position is large, the error in position is integrated to increase the feedback amount, generating a large force to achieve the desired operation (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the prior art, for example, when opening a pull top of canned juice etc. where a large force is required for the operation, if an object that suddenly requires no force at a certain timing is operated by position integral control, it may behave like it runs wild at the moment the force is released, and there is a risk of destroying the object. [[ID=3,7]]

[0005] The present invention has been made in view of the above problems, and an object thereof is to provide an end effector control method capable of smoothly realizing the operation of an object with large temporal variations in work that requires a large operating force.

Means for Solving the Problems

[0006] (1) To achieve the above objective, an end effector control method according to one aspect of the present invention is an end effector control method capable of manipulating an object, wherein the operation is controlled by a hybrid of the joint angle and joint torque of the end effector, and includes the steps of: acquiring the contact point position between the end effector and the target object using a sensor; estimating the force applied to the target object by the end effector based on the acquired contact position; estimating the error between the object operation model when manipulating the target object and the force acting on the target object and the change in the position of the target object; and redetermining the force required for the operation from the estimated amount of error with the object operation model.

[0007] (2) In order to achieve the above objective, an end effector control method according to one aspect of the present invention is an end effector control method capable of manipulating an object, comprising the steps of: estimating the state and required moment of a target object; calculating a torque command and a fingertip force command using a reference fingertip force command and the estimated state and required moment of the target object; estimating a contact point using a contact point estimation instruction and the actual fingertip force detected by a sensor provided on the finger portion of the end effector; controlling fingertip compliance and generating a correction position command using distribution information indicating the distribution of fingertip force based on the estimated contact point and the fingertip force command; and controlling the operation of the end effector by performing impedance control using the correction position command, the torque command, and the fingertip position detected by a sensor provided on the finger portion of the end effector.

[0008] (3) In addition, in an end effector control method according to one aspect of the present invention, the step of estimating the error may be to estimate the angle of the target object based on the fingertip position, generate an inverse model of the motion model of the target object based on the angle of the target object, input the inverse model of the motion model of the target object into a first filter, calculate the torque expected from the movement of the target object based on the output of the first filter, calculate the operating torque of the target object based on the fingertip force, input the operating torque of the target object into a second filter, calculate the torque actually applied to the target object based on the output of the second filter, and calculate the error by dividing the torque expected from the movement of the target object by the torque actually applied to the target object.

[0009] (4) In addition, in an end effector control method according to one aspect of the present invention, an estimated object speed, which is the speed at which the object changes, is estimated based on the change in position information acquired from the sensor, and the limit on the speed of the operation is changed using the estimated object speed. [Effects of the Invention]

[0010] According to (1) to (4), in tasks requiring significant maneuvering force, it is possible to smoothly operate objects that exhibit large temporal fluctuations. [Brief explanation of the drawing]

[0011] [Figure 1] This figure shows an example configuration of a robot hand that performs the tasks according to the embodiment. [Figure 2] This is a diagram illustrating the names of the different parts of a beverage can. [Figure 3] This diagram illustrates an example of operation in each frame during end-effector work. [Figure 4] This figure shows an example of how an end effector works when opening a can's pull tab. [Figure 5]This figure shows an example of how an end effector works when opening a can's pull tab. [Figure 6] This figure shows an example configuration of an end effector control device according to an embodiment. [Figure 7] This figure shows an example of the processing routine when opening a pull tab. [Figure 8] This is a diagram illustrating fingertip force and required moment. [Figure 9] This figure shows an example of processing by the object state estimation unit according to the embodiment. [Figure 10] This is a flowchart of the processing procedure of the end effector control device according to the embodiment. [Figure 11] This figure shows examples of values ​​in the object state estimation unit when the end effector is made to perform the task of opening a pull tab. [Figure 12] This figure shows examples of the state, commands, values, and manipulated quantities during fingertip force control. [Modes for carrying out the invention]

[0012] Embodiments of the present invention will be described below with reference to the drawings. Note that in the drawings used in the following description, the scale of each component has been appropriately changed to ensure that each component is recognizable. In all the figures used to illustrate the embodiments, components with the same function are given the same reference numerals, and repeated explanations are omitted. Furthermore, in this application, "based on XX" means "based on at least XX," and includes cases where it is based on another element in addition to XX. Also, "based on XX" is not limited to cases where XX is used directly, but also includes cases where it is based on something that has been calculated or processed from XX. "XX" is any element (for example, any information).

[0013] <Overview> The end effector of the embodiment estimates the state of the object by estimating the actual contact point position and force using force sensors mounted on the fingertips of the end effector. Then, the end effector estimates the error of the object operation model from the force acting on the object and the estimated motion of the object, and determines the force required to perform the desired motion. Then, the end effector performs robust control against the position error mainly based on torque control by giving a torque command for the above force to the hybrid control of the joint angle and joint torque.

[0014] <Configuration Example of End Effector> FIG. 1 is a diagram showing a configuration example of an end effector that performs work according to the present embodiment. As shown in FIG. 1, the end effector 1 includes finger portions 101, 102, 103, 104, a base body 111, and a camera 131 (sensor). The end effector 1 is connected to an arm 121 via a joint. Further, the finger portion 第101 includes, for example, a force sensor 141 (sensor) at the fingertip. The finger portion 102 includes, for example, a force sensor 142 (sensor) at the fingertip. The finger portion 103 includes, for example, a force sensor 143 (sensor) at the fingertip. The finger portion 104 includes, for example, a force sensor 144 (sensor) at the fingertip. The finger portion 101 corresponds to, for example, the thumb of a human, the finger portion 102 corresponds to, for example, the index finger of a human, the finger portion 103 corresponds to, for example, the middle finger of a human, and the finger portion 104 corresponds to, for example, the ring finger of a human. Note that the end effector 1 includes at least three finger portions. Note that the number of finger portions may be five or more. Further, the end effector 1 includes at least one camera.

[0015] The camera 131 is a CCD (Charge Coupled Device) imaging device, a CMOS (Complementary Metal Oxide Semiconductor) imaging device, or the like. Note that the imaging device 5 may be an RGB-D camera capable of depth measurement. Note that the camera 131 may be installed inside the palm or in the working space.

[0016] <Example of Work Flow> Next, we will explain examples of operations in each frame during work performed by the end effector. For the following explanation, we will use the example of opening a can's pull tab. Figure 2 is a diagram illustrating the names of the different parts of a beverage can. Figure 2 shows an example of a pull-tab can 500. The pull tab 501 comprises a ring-shaped tab 502 that is pulled up by hooking a finger into it, a rivet 503 that secures the tab 502 to the score 504, and the score 504 that is opened. To open the can 500, the ring-shaped tab 502 is pulled by hand and detached from the score 504. When actually drinking the beverage, the pull tab 501 is detached from the can 500, but this is omitted in the following explanation.

[0017] Figure 3 illustrates an example of operation in each frame during end-effector work. In Figure 3, the horizontal direction represents a frame. When the frame is 0, it is the initial state for starting work. Note that the initial state is when the finger portion 102 is pressed against the top surface of the can. In the example in Figure 3, the can is fixed by a fixing device, but if there are two end effectors, the can may be gripped and fixed with the end effector that is different from the one performing the work. When the frame is set to 15, the pull tab 501 lifts up and a hole is made with a "whoosh" sound. Between frames 27 and 28, the score of 504 is cut off in one frame, and the pull tab opens to 501 all at once. When frame 30 is selected, pull tab 501 is fully open. Note that the operation example shown in Figure 3 is just one example and is not limited to this. For example, the timing of operation may differ depending on the shape of the pull tab, etc.

[0018] <Example of how to open a pull tab> Next, we will explain an example of the operation when opening a can's pull tab using an end effector, using Figures 4 and 5. Figures 4 and 5 are diagrams illustrating an example of the operation when opening a can's pull tab using an end effector. In Figures 4 and 5, images g11 to g18 are in chronological order and show the state of each finger, the state of the pull tab, the direction of the applied force, etc. Also, in images g11 to g18, lines g21 (g21-1, g21-2) (red) and g22 (g22-1, g22-2) (blue) represent the coordinate system (e.g., x axis, y axis) of the sensor on the fingertip. Line g23 (g23-1, g23-2) (yellow) represents the actual force being generated. Line g24 (g24-1, g24-2) (purple) represents the force command.

[0019] First, the end effector 1 moves its finger portion closer to the workpiece (g11 in Figure 4). Next, the end effector 1 moves its finger closer to the target, for example, the edge of a can or the pull tab of a pull-top (g12). Image g13 shows the state where finger portion 101 is in contact with the edge of the can and finger portion 102 is in contact with the pull tab. Image g14 shows the state when force is being applied to the pull tab with the finger 102.

[0020] Image g15 in Figure 5 shows the state when the pull tab 501 is lifted by the finger 102, causing the pull tab 501 to lift and the hole to pop open. Image g16 shows the state where the pull tab 501 is opened all at once because the finger 102 further lifts the pull tab 501, causing the score 504 to break. Image g17 shows the pull tab 501 being further lifted by the finger 102, with the pull tab 501 fully open. At this time, the force command applied to the finger 102 is large, as shown by line g24-2. Image g18 shows the pull tab 501 fully open, with the pull tab 501 slightly retracted due to the recoil. Note that the states shown in Figures 4 and 5 are excerpts of the states during the process of opening the pull tab.

[0021] <Example configuration of an end-effector control device> Next, we will describe an example configuration of an end-effector control device. Figure 6 shows an example of the configuration of an end effector control device according to this embodiment. As shown in Figure 6, the end effector control device 2 includes a fingertip movement generation unit 21, a fingertip compliance control unit 22, an object state estimation unit 23, a fingertip force command calculation unit 24, and a speed limit change unit 25.

[0022] The end effector 1 is equipped with force sensors 141-144 on each fingertip and position sensors 151-154 that detect the position of each finger. The force sensors 141-144 are, for example, 6-axis force sensors. The end effector 1 performs work on the object to be operated 3 in accordance with the control of the end effector control device 2. The end effector 1 receives correction position instructions and finger torque commands from the end effector control device 2. The correction position instructions indicate the angle of the finger joints. The end effector 2 outputs to the end effector control device 2 a value indicating the operating force, which is the sensor value detected by the force sensors 141-144, and a value indicating the fingertip position, which is the sensor value detected by the position sensors 151-154.

[0023] The fingertip movement generation unit 21 generates a time-series target position command based on the final target position input from an external device (e.g., a robot control device, a personal computer, etc.) and the modified speed limit information input from the speed limit change unit 25.

[0024] The fingertip compliance control unit 22 controls the end effector 1 by generating a corrected position instruction and a finger torque instruction based on a force instruction that determines the pressing force input from an external device, a time-series target position instruction input from the fingertip trajectory generation unit 21, an estimated object force value input from the object state estimation unit 23, and a fingertip force instruction input from the fingertip force instruction calculation unit 24.

[0025] The object state estimation unit 23 estimates the object force, object resistance force, and object velocity based on the values ​​indicating the operating force detected by the force sensors 141 to 144 and the values ​​indicating the fingertip position detected by the position sensors 151 to 154. In other words, the object state estimation unit 23 outputs to the fingertip force command calculation unit 24 an estimated value of the torque that attempts to open the pull tab due to the force applied by the finger to the pull tab. The resistance force is an estimated value of the lost torque when there is little movement despite the application of the above force.

[0026] The fingertip force command calculation unit 24 calculates a fingertip force command based on the estimated object force and estimated object resistance force input from the object state estimation unit 23, and the force command for determining the pushing force input from an external device.

[0027] The speed limit modification unit 25 modifies the corrected speed limit information based on the estimated speed of the object input from the object state estimation unit 23.

[0028] <Processing routine during operation> Next, we will explain an example of a processing routine when opening a pull tab. Figure 7 shows an example of a processing routine when opening a pull tab.

[0029] The estimation unit 201 estimates the state of the target object and the required moment. The estimation unit 201 corresponds to the fingertip compliance control unit 22.

[0030] The calculation unit 202 calculates the torque command (joint torque) and the fingertip force command using the reference fingertip force command and the estimated value from the estimation unit 21. The calculation unit 202 corresponds to the fingertip force command calculation unit 24.

[0031] The estimation unit 203 estimates the contact point using the actual fingertip force detected by the force sensors provided on the finger portions 101 to 104 of the end effector 1.

[0032] The distribution unit 204 generates distribution information indicating the distribution of fingertip force based on the estimated contact point input from the estimation unit 203.

[0033] The control unit 205 uses the distribution information input from the distribution unit 204 and the fingertip force command input from the calculation unit 202 to control fingertip compliance and generate a correction position command. The control unit 205 corresponds to the fingertip compliance control unit 22.

[0034] The control unit 206 performs impedance control using the correction position command input from the control unit 205, the torque command input from the calculation unit 202, and the fingertip position detected by the position sensors provided on the finger portions 101 to 104 of the end effector 1.

[0035] The fingertip force command is, for example, the force command to the fingertip to open a pull tab, as shown in Figure 8. The required moment is the moment required to lift the pull tab, as also shown in Figure 8. Figure 8 is a diagram illustrating fingertip force and required moment.

[0036] <Processing by the object state estimation unit> Next, an example of the processing of the object state estimation unit 23 will be described. Figure 9 shows an example of the processing of the object state estimation unit according to this embodiment. For the explanation, the example of opening a pull tab will be used.

[0037] The object state estimation unit 23 estimates the angle of the pull tab based on the fingertip position (S231). The object state estimation unit 23 generates an inverse model of the motion model of the pull tab based on the angle of the pull tab (S232). The object state estimation unit 23 inputs the inverse model of the pull tab's motion model to the first filter (S233). The first filter is, for example, a Butterworth filter or an LPF (low-pass filter). The object state estimation unit 23 calculates the torque expected from the movement of the pull tab based on the output of the first filter (S234).

[0038] The object state estimation unit 23 calculates the operating torque of the pull tab based on the fingertip force (S235). The object state estimation unit 23 inputs the operating torque of the pull tab to the second filter (S236). The second filter is, for example, a Butterworth filter or an LPF (low-pass filter). The object state estimation unit 23 calculates the torque actually applied to the pull tab based on the output of the second filter (S237).

[0039] The object state estimation unit 23 estimates the estimated disturbance torque by dividing the torque expected from the movement of the pull tab by the torque actually applied to the pull tab (S238).

[0040] In this embodiment, the motion model of the pull tab corresponds to the object manipulation model. Therefore, the object manipulation model differs for each object being worked on. The end effector control device 2 constructs such object manipulation models in advance, for example, through learning or simulation, using training data and measured data of force, position, and commands during operation, and stores the constructed model.

[0041] <Processing procedure for the end effector control device> Next, an example of the processing procedure of the end effector control device 2 will be described. Figure 10 is a flowchart of the processing procedure of the end effector control device according to this embodiment.

[0042] (Step S301) The object state estimation unit 23 acquires the image captured by the camera 131, the sensor values ​​detected by the force sensors 141 to 144, and the position information detected by the position sensors 151 to 154.

[0043] (Step S302) The object state estimation unit 23 detects the position of the target object and the fingertip position of the end effector 1 based on the image and sensor values ​​acquired from the sensor. The estimation unit 203 estimates the contact point using the contact point estimation instruction input from the external device and the actual fingertip force detected by the force sensors provided on the finger parts 101 to 104 of the end effector 1.

[0044] (Step S303) The object state estimation unit 23 estimates the object force and object resistance force based on the sensor values ​​obtained from the sensor. In this way, the object state estimation unit 23 and the fingertip force calculation unit 24 estimate the force applied to the object by the end effector 1 based on the contact position between the end effector 1 and the object.

[0045] (Step S304) As shown in Figure 9, the object state estimation unit 23 estimates the error (estimated disturbance torque) between the object operation model when operating the object and the force acting on the object and the change in the position of the object.

[0046] (Step S305) The fingertip position command calculation unit 24 and the fingertip compliance difference control unit 22 re-determine the force required for the operation based on the estimated amount of error with the object operation model.

[0047] <Example of actual measurement data> Next, we will explain examples of how the values ​​change when the end effector 1 is made to open the pull tab using the method of this embodiment. Figure 11 shows examples of values ​​in the object state estimation unit when the end effector is made to open the pull tab. The horizontal axis is the frame, the vertical axis of graph g100 is the estimated force, and the vertical axis of graph g150 is the control values ​​for angle, angular velocity, and velocity limit. Period T1 is the initial state, period T2 is when the pull tab opens with a "whoosh," and period T3 is when the pull tab opens all at once.

[0048] In graph g100, line g101 represents the change in torque of the pull tab, line g102 represents the change in torque estimated from the angular velocity of the pull tab, line g103 represents the change in the value of the torque of the pull tab after passing it through a filter, and line g104 represents the change in the disturbance estimate.

[0049] In graph g150, line g151 represents the change in the angle of the pull tab, line g152 represents the change in the angular velocity of the pull tab, and line g153 represents the change in the velocity limit.

[0050] As shown in graph g100, the torque of the pull tab decreases sharply when it opens all at once, then increases again, and then decreases again. The torque estimated from the angular velocity of the pull tab increases gradually from the initial state until the pull tab opens with a "whoosh," remains almost constant from the "whoosh" until it opens all at once, and then increases when it opens all at once. The value after passing through the filter increases gradually from the initial state until after it opens all at once. Also, the disturbance estimate decreases from before the pull tab opens with a "whoosh" until it opens all at once, and then increases slightly after it opens all at once.

[0051] Furthermore, as shown in graph g150, the angle and angular velocity of the pull tab increase sharply when the pull tab is opened all at once, and the angle of the pull tab then returns slightly from the open position. The angular velocity of the pull tab decreases sharply after it opens all at once. Note that, as shown in graph g150, the speed limit was not changed when the pull tab was opened in this experiment.

[0052] Figure 12 shows examples of the state, command, values, and manipulated amount during fingertip force control. The horizontal axis represents the frame, the vertical axis of graph g200 represents the change in state, the vertical axis of graphs g210 and g230 represents force (N), and the vertical axis of graphs g220 and g240 represents the manipulated amount (m).

[0053] The line g201 in graph g200 represents the change in the operating state. In graph g210, line g211 represents the change in the actual force acting in the X-axis direction, line g212 represents the change in the force command in the X-axis direction, and line g213 represents the change in the filter value of the force command in the X-axis direction. In graph g220, line g221 represents the change in the compliance control variable in the X-axis direction, and line g222 represents the change in the compliance integral control variable in the X-axis direction. In graph g230, line g231 represents the change in the actual force acting in the Z-axis direction, line g232 represents the change in the force command in the Z-axis direction, and line g233 represents the filter value of the force command in the Z-axis direction. In graph g240, line g241 represents the change in the compliance control variable in the Z-axis direction, and line g242 represents the change in the compliance integral control variable in the Z-axis direction.

[0054] As shown in graph g200, the initial state and the period when the pull tab opens all at once represent the phase where force is applied to open the pull tab, while the phase after it opens all at once represents the phase where the force is released. As shown in graph g210, when the operation was performed using the control method of this embodiment, the force actually acting on the force command was able to follow in the X-axis direction. Furthermore, as shown in graph g230, when performing operations using the control method of this embodiment, the force actually acting on the force command is able to follow the Z-axis direction, although not as consistently as in the X-axis direction.

[0055] Note that the states, commands, manipulated variables, forces, etc. shown in Figures 11 and 12 are examples only and are not limited to these.

[0056] In this embodiment, the state of the object is estimated by estimating the actual contact point position and force using a force sensor mounted on the fingertip. Furthermore, in this embodiment, the force required to perform the desired action is determined by estimating the error (estimated disturbance torque) in the object operation model from the force acting on the object and the estimated movement of the object. In addition, in this embodiment, a torque command to generate the above force is applied to a hybrid control of joint angle and joint torque, thereby enabling control that is robust to position errors and primarily relies on torque control.

[0057] As a result, according to this embodiment, it is possible to smoothly perform operations on objects that require a large operating force, such as opening a pull tab, but whose operating force varies greatly over time.

[0058] In the example described above, opening a pull tab was explained as an example of a task requiring significant force, but this is not the only example. Such tasks include, for example, pulling and cutting paper or rubber, drilling holes, and pushing out carts, etc. The method of this embodiment can also be applied to these tasks.

[0059] Furthermore, the end effector 1 that performs the work may be provided by a robot. In such a case, the robot may be a bipedal robot equipped with two end effectors. When there are two end effectors, the end effector control device 2 determines a taxonomy (see Reference 1) indicating the work content and gripping state based on, for example, a captured image and operation instructions from the operator. The end effector control device 2 may then control each end effector using the method described above, based on the taxonomy and instructions from the operator. Note that there may be three or more end effectors 1. In this case as well, the end effector control device 2 may then control each end effector using the method described above, based on the taxonomy and instructions from the operator.

[0060] Reference 1; Thomas Feix, Javier Romero, et al., “The GRASP Taxonomy of Human Grasp Types” IEEE Transactions on Human-Machine Systems (Volume: 46, Issue: 1, Feb. 2016), IEEE, p66-77

[0061] Furthermore, a program to implement all or part of the functions of the end effector control device 1 in this invention may be recorded on a computer-readable recording medium, and all or part of the processing performed by the end effector control device 1 may be performed by loading the program recorded on this recording medium into a computer system and executing it. Herein, "computer system" includes hardware such as an OS and peripheral devices. Furthermore, "computer system" also includes a WWW system equipped with a homepage provisioning environment (or display environment). Furthermore, "computer-readable recording medium" refers to portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, and storage devices such as hard disks built into a computer system. Moreover, "computer-readable recording medium" also includes volatile memory (RAM) inside a computer system that acts as a server or client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line, which holds the program for a certain period of time.

[0062] Furthermore, the above program may be transmitted from a computer system that stores the program in a memory device or the like to another computer system via a transmission medium or by transmission waves within the transmission medium. Here, the "transmission medium" for transmitting the program refers to a medium that has the function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. In addition, the above program may be for the purpose of realizing a part of the functions described above. Furthermore, it may be a so-called differential file (differential program) that can realize the functions described above in combination with a program already recorded in the computer system.

[0063] Although embodiments for carrying out the present invention have been described above using examples, the present invention is not limited in any way to these embodiments, and various modifications and substitutions can be made without departing from the spirit of the present invention. [Explanation of Symbols]

[0064] 1...End effector, 101, 102, 103, 104...Finger part, 111...Base, 131...Camera, 121...Arm, 141, 142, 143, 144...Force sensor, 151, 152, 153, 154...Position sensor, 2...End effector control device, 21...Fingertip movement generation unit, 22...Fingertip compliance control unit, 23...Object state estimation unit, 24...Fingertip force command calculation unit, 25...Speed ​​limit change unit, 201...Estimation unit, 202...Calculation unit, 203...Estimation unit, 204...Distribution unit, 205...Control unit, 206...Control unit

Claims

1. An end effector control method that allows manipulation of an object, The operation is controlled by a hybrid of the joint angle and joint torque of the end effector. The sensor acquires the contact position between the end effector and the target object. A step of estimating the force applied to the target object from the end effector based on the acquired contact position, A step of estimating the error between the object manipulation model when manipulating the object and the force acting on the object and the change in the position of the object, The process involves re-determining the force required for the operation based on the estimated error amount with respect to the aforementioned object manipulation model, An end effector control method having [a specific feature].

2. An end effector control method that allows manipulation of an object, A process for estimating the state of the target object and the required moment, A process for calculating torque commands and fingertip force commands using a reference fingertip force command, the estimated state of the target object, and the required moment, The process involves estimating the contact point using the actual fingertip force detected by the sensor on the finger portion of the end effector, A step of controlling fingertip compliance and generating a corrected position command using distribution information indicating the distribution of fingertip force based on the estimated contact point and the fingertip force command, A step of controlling the operation of the end effector by performing impedance control using the correction position command, the torque command, and the fingertip position detected by the sensor provided on the finger portion of the end effector, An end effector control method having [a specific feature].

3. The process of estimating the aforementioned error is as follows: The angle of the target object is estimated based on the fingertip position detected by the sensor on the finger portion of the end effector. Based on the angle of the object, an inverse model of the object's motion is generated. The inverse model of the motion model of the target object is input to the first filter. Based on the output of the first filter, the torque expected from the movement of the target object is calculated. Based on the fingertip force, the operating torque of the target object is calculated. The operating torque of the target object is input to the second filter. Based on the output of the second filter, the torque actually applied to the target object is calculated. The error is calculated by dividing the torque expected from the movement of the object by the torque actually applied to the object. The end effector control method according to claim 1.

4. Based on the change in position information obtained from the aforementioned sensor, the estimated velocity of the target object, which is the rate at which the target object changes, is estimated. Using the estimated object velocity, the limit on the speed of the operation is changed. The end effector control method according to claim 1 or claim 3.

Images