Torque estimation method and torque estimation device

Abstract

This torque estimation method involves taking the total sum of moments at all of a plurality of measurement points at each measurement point as obtained by multiplying the moment arm from a reference point to the measurement points and the divergence of displacement vectors of the measurement points displaced by an external force on a facing surface that faces an object, and estimating the torque in a direction inclined with respect to the facing surface.

Description

Torque estimation method and torque estimation device

[0001] The present disclosure relates to a torque estimation method and a torque estimation device.

[0002] Zhang, Yazhan, et al. "Effective estimation of contact force and torque for vision-based tactile sensors with helmholtz-hodge decomposition." IEEE Robotics and Automation Letters 4.4(2019): 4094-4101. describes a technology that uses an image-based contact sensor with multiple measurement points on the surface facing an object, and estimates the force in the normal direction of the opposing surface by taking the divergence of the displacement vector at the measurement points.

[0003] The method described in the above document can estimate the force in the normal direction of the opposing surface of the image-type contact sensor, but there is room for improvement in terms of more accurate torque estimation.

[0004] The present disclosure aims to improve the accuracy of torque estimation using an image-type contact sensor.

[0005] One aspect of the present disclosure is a torque estimation method that estimates torque in a direction inclined with respect to a normal to a surface facing an object by taking the sum of moments at all measurement points, which are obtained by multiplying the divergence of displacement vectors at multiple measurement points displaced by an external force on the surface facing the object by the moment arm from a reference point to the measurement points.

[0006] According to the present disclosure, it is possible to improve the accuracy of torque estimation using an image-type contact sensor.

[0007] FIG. 1 is a front view of an end effector equipped with a torque estimation device to which the torque estimation method of the first embodiment is applied. FIG. 2 is a perspective view showing an example of an image-based contact sensor used in the torque estimation method of the first embodiment. FIG. 3 is a configuration diagram showing a computer that executes the torque estimation method of the first embodiment. FIG. 4 is a flowchart showing the calculation processing in the torque estimation method of the first embodiment. FIG. 5A is a distribution diagram showing the force on the facing surface of the image-based contact sensor when an object is not being grasped. FIG. 5B is a distribution diagram showing the force on the facing surface of the image-based contact sensor when an object is being grasped. FIG. 5C is a distribution diagram showing the torque on the facing surface of the image-based contact sensor when an object is being grasped, before zero adjustment. FIG. 5D is a distribution diagram showing the torque on the facing surface of the image-based contact sensor when an object is being grasped, after zero adjustment.

[0008] A torque estimation method and a torque estimation device 12 according to an embodiment of the present disclosure will be described below with reference to the drawings, along with an end effector 14 to which the torque estimation method is applied. The same reference numerals are used throughout the drawings to designate identical or substantially equivalent elements, members, and parts. The dimensions and proportions of the drawings are exaggerated for the sake of clarity and may differ from the actual proportions.

[0009] 1, the end effector 14 has a pair of gripping portions 14A and 14B. The gripping portions 14A and 14B are adapted to move toward and away from each other by, for example, an actuator 16. When the gripping portions 14A and 14B move toward each other, they can grip an object BJ.

[0010] An image-type contact sensor 18 is attached to one of the gripping portions 14A. The image-type contact sensor 18 is an example of a tactile sensor of the disclosed technology. As also shown in FIG. 2 , the image-type contact sensor 18 has a housing 20. One surface of the housing 20 is a facing surface 22. In the illustrated example, the facing surface 22 is rectangular. Hereinafter, the horizontal direction, vertical direction, and normal direction of the facing surface 22 will be referred to as the X-axis direction, Y-axis direction, and Z-axis direction, respectively.

[0011] An image-type contact sensor 18 is attached to one of the gripping parts 14A. The image-type contact sensor 18 is oriented such that the facing surface 22 faces the object BJ. In contrast, in the example shown in Fig. 1, the other gripping part 14B does not have the image-type contact sensor 18 attached, but instead has a housing 20 of the same shape attached. However, the other gripping part 14B may also have the image-type contact sensor 18 attached.

[0012] A plurality of measurement points 24 are set on the opposing surface 22. In the example shown in Fig. 2, a predetermined number of measurement points 24 are arranged in the X-axis direction, and a predetermined number (which may be different from the number in the X-axis direction) are also arranged in the Y-axis direction, so that the measurement points 24 are arranged in a matrix pattern as a whole. A camera (not shown) is arranged inside the image-type contact sensor 18, which can capture the positions of the measurement points 24 as an image and obtain the amount of displacement.

[0013] 3 shows the configuration of a torque estimation device 12 that applies the torque estimation method. The torque estimation device 12 includes the image-type contact sensor 18 and a computer 40. The computer 40 is an example of a computing device of the disclosed technology.

[0014] The computer 40 comprises a processor 42, memory 44, storage 46, an input device 48, an output device 50, a storage medium reading device 52, and an interface (I / F) 54, and each of these elements is communicatively connected to each other by a bus 56.

[0015] The storage 46 stores a calculation processing program 58 for executing calculation processing of the torque estimation method. The processor 42 is capable of executing various programs and controlling each element. Specifically, the processor 42 reads the program from the storage 46 and executes the program using the memory 44 as a work area. In other words, the processor 42 controls each element and performs various calculation processing in accordance with the program stored in the storage 46.

[0016] The memory 44 serves as a working area and can temporarily store programs and various data. The storage 46 is, for example, a read-only memory (ROM), a hard disk drive (HDD), or a solid state drive (SSD), and stores various programs and data. These programs include not only application programs such as the above-mentioned arithmetic processing program, but also an operating system.

[0017] The input device 48 is a device for providing various inputs to the computer 40. The input device 48 may include operation switches, operation buttons, etc., as well as pointing devices such as a keyboard and a mouse used in personal computers, etc. The output device 50 is a device for outputting various information from the computer 40, and may include, for example, an indicator lamp, a speaker, etc. in addition to a display. A touch panel display may be used as a display, which is an example of the output device 50. In this case, the touch panel display also functions as the input device 48.

[0018] The storage medium reader 52 is a device that reads data stored in various storage media and writes data to the storage media. Examples of storage media include CD (Compact Disc)-ROM, DVD (Digital Versatile Disc)-ROM, Blu-ray Disc, and USB (Universal Serial Bus) memory. The interface 54 is a device for communicating with other devices. For communication, standards such as Ethernet (registered trademark) and FDDI (Fiber Distributed Data Interface) are used.

[0019] The torque estimation device 12 performs the torque estimation method of the disclosed technology by causing the computer 40 to execute a calculation processing program according to the flow shown below, for example. This calculation processing program is executed, for example, when the end effector 14 is gripping the object BJ.

[0020] First, in step S102, the computer 40 calculates the displacement vector v of each measurement point 24 on the opposing surface 22. ｉ This displacement vector v ｉ is a vector having components in the X-axis direction and the Y-axis direction at each of the plurality of measurement points 24. ｉ may be obtained as a change in the position of a marker set at each of the measurement points 24, or may be obtained by calculating a change in the optical flow obtained from visual information of the measurement points 24. In either case, the computer 40 calculates the displacement vector v ｉ is acquired based on the movement of the markers set at the respective measurement points 24. Here, the number of measurement points 24 is set to N, and these N measurement points 24 are numbered i (1≦i≦N) to represent v ｉ It is distinguished as:

[0021] Next, in step S104, the computer 40 calculates the displacement vector v ｉ Specifically, as shown in equation (1), we take the inner product with the differential operator ∇.

[0022]

[0023] FIG. 5B shows a displacement vector v as a vector field on the opposing surface 22 after the end effector 14 has grasped the object BJ. ｉ 5B, the positions indicated by "+" and "-" are the centers of gravity of the positive and negative portions of the divergence components, and the divergence components gradually decrease with increasing distance from these positions. FIG. 5C also shows the distribution of the divergence components when the end effector 14 is gripping the object BJ. In the example shown in FIG. 5C, the positions indicated by "+" and "-", which are the positions showing the maximum values ​​of the divergence components, have moved to the upper and lower portions of the opposing surface 22, respectively.

[0024] Next, in step S106, the computer 40 performs zero correction of the divergent components of the opposing surface 22. Specifically, for the torque distribution shown in FIG. 5C, the displacement vector v at the measurement point 24 is corrected to zero so that the sum of the divergent components becomes zero. ｉA fixed correction value is added to or subtracted from the value of .times. ...

[0025] At the "+" position at the top of Figure 5D, the divergence component has a maximum value (maximum absolute value as a positive value), whereas at the "-" position at the bottom, the divergence component has a minimum value (maximum absolute value as a negative value). The vector field shown in Figure 5D resembles the electric field E generated by an electric dipole as a whole, with positive charges at the "+" positions and negative charges at the "-" positions. In other words, when zero correction is performed as shown in Figure 5D, a dipole moment field similar to an electric dipole moment is generated.

[0026] Next, in step S108, the computer 40 calculates the moment arm at each measurement point 24. In calculating the moment arm, first, the divergence set value p is calculated on the positive side and the negative side of the divergence component by the following equations (2) and (3): ｉ ＋ , p ｉ － Set.

[0027]

[0028] Specifically, for example, in the divergence of a vector field, the displacement vector v ｉ The divergence of the positive value of is taken, and the negative part is masked as 0 (zero). Similarly, the divergence setting value of the negative part is the displacement vector v ｉ The divergence of negative values ​​is taken and the positive part is masked as 0 (zero).

[0029] And the divergence setting value p ｉ ＋ , p ｉ － For example, the center of gravity r on the positive side in the X-axis direction is calculated as follows: ｇｘ ＋ Then, as shown in equation (4),

[0030]

[0031] The center of gravity r on the positive side in the Y-axis direction ｇｙ ＋ , the center of gravity r on the negative side in the X-axis direction ｇｘ － , and the center of gravity r on the negative side in the Y-axis direction ｇｙ － The same can be obtained for

[0032] From the position of the center of gravity thus obtained, the coordinate r of the origin of the moment arm is ｇ is calculated by equation (5).

[0033]

[0034] Then, for each measurement point 24, the moment arm (r ｉ ) ｘ , moment arm in the Y-axis direction (r ｉ ) ｙ are calculated using equation (6).

[0035]

[0036] That is, the moment arm is the distance from the reference point, which is the position of the center of gravity of the divergence calculated by equation (5), to each measurement point 24 .

[0037] However, this reference point is not limited to a point determined by the above calculation, and may be, for example, any predetermined point. For the image-type contact sensor 18, an example of such an arbitrary point is the center or edge of the opposing surface 22. Using a point determined by the above calculation as the reference point allows for more accurate torque estimation than using any arbitrary point as the reference point.

[0038] The reference point can be set, for example, at an initial stage of executing the torque estimation method of the disclosed technology. A more specific example is when the torque estimation device 12 is powered on. The reference point may also be reset by recalculating it in response to a predetermined trigger. Examples of the predetermined trigger include a regular cycle or when the torque estimated by the torque estimation method of the disclosed technology is equal to or greater than a predetermined threshold. In particular, when the object BJ moves, the estimated torque may be equal to or greater than the predetermined threshold. When the position of the reference point is static, the reference point does not need to be reset, making it easy to calculate the moment arm. On the other hand, when the position of the reference point is dynamic, it is possible to calculate the moment arm more appropriately in accordance with the movement of the object BJ.

[0039] Here, we define "tactile dipole moment." As described above, in the technology of the present disclosure, the vector field on the facing surface 22 is similar to the electric field E generated by an electric dipole. For this electric field E, if the distance from the reference point is r and the dielectric constant of vacuum is ε0, the electric dipole moment p can be obtained using the differential operator ∇ according to the following equation (7).

[0040]

[0041] In the technique of the present disclosure, in the same manner as when the electric dipole moment p is obtained in this manner, the "tactile dipole moment" is calculated in step S110. tilt is defined by the following equations (8) and (9).

[0042]

[0043] That is, the tactile dipole moment p tilt is calculated by multiplying the divergence of the displacement at each measurement point 24 by the moment arm, and then summing these values ​​for all measurement points 24. This tactile dipole moment p tilt is also a rotational force having two directional components in the X-axis direction and the Y-axis direction.

[0044] Here, the electric dipole moment p tiltIn the case of an electric dipole moment, the direction of this moment is from the negative charge to the positive charge. As in the case of an electric dipole moment, the tactile dipole moment p tilt In step S112, the computer 40 multiplies the tactile dipole moment by a predetermined coefficient to calculate the torque τ tilt That is, in this embodiment, the tactile dipole moment p tilt By obtaining the torque τ in the two component directions in the opposing surface 22, tilt can be estimated.

[0045] In this manner, in this embodiment, the torque τ when the end effector 14 grasps the object BJ is ｔｉｌｔ can be estimated in two directions (X-axis direction and Y-axis direction) within the plane. That is, in this embodiment, the torque when the end effector 14 grips the object BJ can be obtained in the X-axis direction and the Y-axis direction in addition to the Z-axis direction. This allows for more accurate estimation of the torque in a direction inclined relative to the normal to the opposing surface 22 compared to estimating the torque only in the normal direction (Z-axis direction) of the opposing surface 22. This also makes it possible to control the position, posture, etc. of the object BJ with high precision. For example, as shown in FIG. 1 , the object BJ may have an insertion portion BJ-S, and this insertion portion BJ-S may be inserted into a insertion port CJ-S of a mating member CJ. In such a case, the insertion portion BJ-S of the object BJ can be accurately positioned and its posture controlled relative to the insertion port CJ-S.

[0046] In the above example, as shown in Fig. 5D, so-called zero correction is performed on the divergent component of the opposing surface 22. However, even without such zero correction, it is possible to estimate torque in two different directions within the opposing surface 22. However, without zero correction, it is difficult to obtain an appropriate value for the tactile dipole moment when there is a large bias in the divergent component of the opposing surface 22. Therefore, performing zero correction enables torque to be estimated with higher accuracy than not performing zero correction.

[0047] In the above example, the two different directions within the opposing surface 22 are the X-axis direction and the Y-axis direction, which are orthogonal to the Z-axis and orthogonal to each other, but these "two different directions" are not limited to two linearly independent directions within the opposing surface 22. For example, the X-axis and the Y-axis may be two directions that form oblique coordinates.

[0048] Furthermore, in the above example, an image-type contact sensor 18 that acquires images of the positions of multiple measurement points on the facing surface 22 is given as an example of a tactile sensor of the disclosed technology. The tactile sensor is not limited to this type of image-type contact sensor, as long as it is possible to acquire displacement vectors at multiple measurement points 24 on the facing surface 22. For example, elements that displace in the direction of pressure may be placed at the positions of the multiple measurement points 24, and the positions of these elements may be read electrically.

[0049] Although the above example illustrates an application of the technology of the present disclosure to torque estimation when an end effector 14 of a robot or the like grasps an object, application examples of the technology of the present disclosure are not limited to this. For example, the technology can also be applied to devices that estimate torque (contact torque) received from an object. Examples of such devices include standalone sensors that are not robots, such as haptic devices.

[0050] The following are supplementary notes related to the present disclosure. (Supplementary Note 1) A torque estimation method that estimates a torque in a direction inclined with respect to a normal to the opposing surface by taking the sum of moments at all measurement points, which are obtained by multiplying the divergence of displacement vectors at a plurality of measurement points displaced by an external force on an opposing surface facing an object by the moment arm from a reference point to the measurement points. (Supplementary Note 2) The torque estimation method according to Supplementary Note 1, in which a certain correction value is added to or subtracted from the value of the displacement vector at each measurement point so that the sum of the displacement vectors on the opposing surface becomes zero. (Supplementary Note 3) The torque estimation method according to Supplementary Note 1 or Supplementary Note 2, in which the reference point is the center of gravity of the divergence of the displacement vectors on the opposing surface. (Supplementary Note 4) The torque estimation method according to Supplementary Note 3, in which the position of the center of gravity of the divergence of the displacement vectors is determined from the positions of the centers of gravity obtained from the positive and negative parts of the divergence of the displacement vectors. (Supplementary Note 5) The torque estimation method according to any one of Supplementary Notes 1 to 4, wherein the reference point is reset in response to a predetermined trigger. (Supplementary Note 6) The torque estimation method according to any one of Supplementary Notes 1 to 5, wherein the displacement vector is acquired based on movement of markers at the plurality of measurement points set on a tactile sensor. (Supplementary Note 7) A torque estimation device comprising: a tactile sensor having a plurality of measurement points on a facing surface facing an object; and a calculation device that takes the sum, at all measurement points, of moments at the measurement points obtained by multiplying the divergence of displacement vectors at the plurality of measurement points by the moment arm from the reference point to the measurement points, to estimate a torque in a direction inclined with respect to the normal to the facing surface. (Supplementary Note 8) The torque estimation device according to Supplementary Note 7, wherein the tactile sensor acquires the displacement vectors at the measurement points as images.

[0051] The disclosure of Japanese Patent Application No. 2024-6281, filed on January 18, 2024, is incorporated herein by reference in its entirety. All documents, patent applications, and technical standards mentioned herein are incorporated herein by reference to the same extent as if each individual document, patent application, and technical standard was specifically and individually indicated to be incorporated by reference.

Claims

1. A torque estimation method for estimating a torque in a direction inclined with respect to the normal of the facing surface by taking a sum of all moments at the measurement points, which are obtained by multiplying the divergence of displacement vectors of a plurality of measurement points that are displaced by an external force on a facing surface facing an object, and the moment arm from a reference point to the measurement point.

2. The torque estimation method according to claim 1, wherein a constant correction value is added to or subtracted from the value of the displacement vector of each measurement point so that the sum of the displacement vectors on the facing surface becomes zero.

3. The torque estimation method according to claim 1, wherein the center of gravity of the divergence of the displacement vectors on the facing surface is used as the reference point.

4. The torque estimation method according to claim 3, wherein the position of the center of gravity of the divergence of the displacement vectors is obtained from the positions of the centers of gravity respectively obtained from the positive part and the negative part of the divergence of the displacement vectors.

5. The torque estimation method according to claim 1, wherein the reference point is reset from a predetermined trigger.

6. The torque estimation method according to claim 1, wherein the displacement vector is acquired based on the movement of markers at a plurality of the measurement points set on a tactile sensor.

7. A torque estimation device having: a tactile sensor provided with a plurality of measurement points on a facing surface facing an object; and an arithmetic unit that estimates a torque in a direction inclined with respect to the normal of the facing surface by taking a sum of all moments at the measurement points, which are obtained by multiplying the divergence of displacement vectors at the plurality of measurement points and the moment arm from a reference point to the measurement point.

8. The torque estimation device according to claim 7, wherein the tactile sensor acquires the displacement vector of the measurement point as an image.

Images