GRIP POSITION DETERMINATION DEVICE, GRIP POSITION DETERMINATION SYSTEM, GRIP POSITION DETERMINATION METHOD, AND PROGRAM
The gripping position determination device estimates frictional force distribution to determine stable gripping points, addressing slipping issues and surface curvature challenges, ensuring robust and efficient grasping.
Patent Information
- Application Number
- JP2021060662
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2021-03-31
- Publication Date
- 2024-11-11
- Estimated Expiration
- 2041-03-31
AI Technical Summary
Conventional gripping technologies fail to accurately determine stable gripping points due to inconsistencies between general-purpose gripping characteristics and actual finger movements, leading to potential slipping and inability to handle curved surfaces, and lack flexibility in fingertip contact positions.
A gripping position determination device and method that estimates frictional force distribution using predictive control, calculates stable gripping positions, and selects grippable areas considering frictional forces and external forces, allowing for robust gripping even on curved surfaces.
Enables accurate determination of gripping points that minimize slipping and maintain object posture, reducing calculation costs and ensuring stable grasping despite shape or control errors.
Smart Images

Figure 0007580326000001 
Figure 0007580326000002 
Figure 0007580326000003
Abstract
Description
[Technical field]
[0001] The present invention relates to a gripping position determining device, a gripping position determining system, a gripping position determining method, and a program. [Background technology]
[0002] A control device that allows a robot to grasp an object has been proposed. As such a control device, a method has been proposed in which a grasping point that can be physically contacted is tentatively determined, a grasping force is determined taking into consideration the balance of forces, the quality of the grasping force is evaluated as the volume of the envelope of the grasping force, and a grasping point with higher quality is searched for (for example, see Patent Document 1).
[0003] In addition, in the technology described in Patent Document 2, before actually gripping an object, only the middle finger is slid on the surface of the object to be gripped with a predetermined pressure, the pressure and frictional force at that time are detected, and an estimated friction coefficient is calculated from the ratio of the pressure and frictional force. Then, in the technology described in Patent Document 2, the gripping force is maintained so as not to cause slippage using the calculated estimated friction coefficient, while the gripping force is appropriately controlled so as not to become excessive, thereby stably holding the object with an appropriate gripping force.
[0004] FIG. 13 is a diagram for explaining an example in which a robot grasps an object (coin-shaped object Obj) with two fingers (upper finger 901 and lower finger 902) in the conventional technology. In the first state g901, the inter-finger gap, which is the deviation in the longitudinal direction of the coin Obj between the upper finger 901 and the lower finger 902, is 0.0 (m). In the second state g902, the inter-finger gap between the upper finger 901 and the lower finger 902 is 0.0015 (m). The graph g903 is a graph showing the relationship between the inter-finger gap and the lateral force (sum of squares) including the first state and the second state, where the horizontal axis is the inter-finger gap (m) and the vertical axis is the lateral force. The lateral force is a lateral force (friction force) applied to the grasped object. As shown in FIG. 13, as the inter-finger gap increases, the lateral force increases and the grasped state becomes unstable.
[0005] FIG. 14 is a diagram for explaining an example in which a robot in the prior art grasps an object (spherical object Obj) with three fingers (first finger 911, second finger 912, and third finger 913). Graphs g911 to g914 represent the first to fourth states. In the graphs g911 to g914, the vertical and horizontal axes represent the inter-finger gap (m). In the first state as shown in graph g911, the angles of the three fingers are 10, 170, and 270 degrees, respectively, and the lateral force is 6.21e -23 In the second state, as shown in graph g912, the angles of the three fingers are 10, 170, and 230 degrees, respectively, and the lateral force is 9.81e -2 When the fingers are in a position opposite to gravity, as in graphs g911 and g912, the lateral force is at a minimum. In the third state, as in graph g913, the angles of the three fingers are 30, 150, and 270 degrees, respectively, and the lateral force is 6.29e -23 In the fourth state, as shown in graph g914, the angles of the three fingers are 70, 150, and 270 degrees, respectively, and the lateral force is 5.72e -07 As shown in graphs g913 and g914, the lateral force is minimized when the fingers are positioned symmetrically.
[0006] FIG. 15 is a diagram for explaining an example of the index ε (see Non-Patent Document 1) when a robot in the prior art grasps an object (coin-shaped object Obj) with two fingers (upper finger 901 and lower finger 902) with an inter-finger gap of 0.0 (m). Graph g921 is a graph showing the inter-finger gap. Graph g922 is a diagram showing the index ε in two dimensions, with the horizontal axis representing force and the vertical axis representing moment. The larger the area surrounded by the dashed line, the more stable the grasping state. Thus, the index ε when the fingers face each other in the thickness direction of the object (inter-finger gap 0.0 (m)) is about 5.69e -06 Graph g923 shows the friction and force vector of the upper finger, and graph g924 shows the friction and force vector of the lower finger. In this state, G c f c =-F e is established, and there is sufficient room for grasping. e is the external force, G c f cis the force that the robot may exert when grasping. The condition for force closure is that any external force F e ∈R 6 For G c f c =-F e is true.
[0007] FIG. 16 is a diagram for explaining an example of an index ε (see Non-Patent Document 1) when a robot in the prior art grasps an object (coin-shaped object Obj) with two fingers (upper finger 901 and lower finger 902) with an inter-finger gap of 0.0015 (m). Graph g931 is a graph showing the inter-finger gap. Graph g932 is a diagram showing the index ε in two dimensions. In this way, the index ε when the inter-finger gap of 0.0015 (m) is shifted is approximately 1.21e -05 Graph g933 shows the friction and force vector of the upper finger, and graph g934 shows the friction and force vector of the lower finger. In this state, G c f c =-F e is established, but there is little room for grasping. [Prior art documents] [Patent documents]
[0008] [Patent Document 1] Patent No. 6476358 [Patent Document 2] JP 2005-144573 A [Non-patent literature]
[0009] [Non-Patent Document 1] Andrew T. Miller Peter K. Allen, “GraspIt!: A Versatile Simulator for Grasp Analysis”, IEEE Robotics & Automation Magazine (Volume: 11, Issue: 4, Dec. 2004),2004,p110-122 Summary of the Invention [Problem to be solved by the invention]
[0010] However, in the conventional technology, when moving only a specific finger along the surface of an object with the same number of fingers in contact, the index ε and volume v, which are general-purpose gripping characteristics, do not necessarily directly relate to the slipperiness of the finger, and the finger may slip. As a result, the evaluation of index ε may erroneously recognize an unstable state as a stable state, as shown in Figure 16. In addition, in the technology described in Patent Document 2, the friction coefficient is estimated by sliding only the finger on the surface of the object to be grasped with a certain pressure before actually grasping the object, but the contact position of the fingertip tip is restricted for position correction, and there is no degree of freedom. In the technology described in Patent Document 2, for example, the friction coefficient cannot be estimated when the surface of the object is curved. As described above, in the conventional technology, it was difficult to determine a gripping point while taking into account an operation.
[0011] The present invention has been made in consideration of the above-mentioned problems, and aims to provide a grip position determination device, a grip position determination system, a grip position determination method, and a program that can determine a grip point taking into account the operation. [Means for solving the problem]
[0012] (1) In order to achieve the above-mentioned object, a grasping position determination device according to one embodiment of the present invention is a grasping position determination device for a robot hand having multiple fingers with multiple joints, and includes a friction force distribution calculation unit that estimates a friction force between one of the fingers to be grasped and an object from predictive control of a grasping force when an object is grasped with at least two of the fingers, and calculates a friction force distribution at which the object can be grasped on the object surface based on a value related to the friction force calculated using the estimated friction force, a graspable area selection unit that selects at least one graspable area from the friction force distribution, and a grasp position calculation unit that calculates a grasp position at which the object can be stably grasped from the selected graspable area.
[0013] (2) In addition, in a gripping position determination device according to one aspect of the present invention, the frictional force distribution calculation unit may calculate the frictional force distribution within a predetermined range from an initial gripping position at which the object is gripped by one of the fingers to be gripped, and the gripping position calculation unit may search within the frictional force distribution from the initial gripping position to calculate a gripping position at which the object can be gripped stably against an external force.
[0014] (3) Furthermore, in the gripping position determination device according to one aspect of the present invention, the gripping position calculation unit may select the grippable area in which the posture or movement of the object can be maintained in a specific state, and calculate a gripping position within the grippable area in which the posture or movement of the selected object can be maintained in a specific state.
[0015] (4) In addition, in the gripping position determination device according to one aspect of the present invention, the frictional force distribution calculation unit may calculate a force that the robot hand is permitted to exert, and estimate the frictional force by solving a balance equation between the calculated force that the robot hand is permitted to exert and an external force predicted when gripping the object.
[0016] (5) In addition, in the gripping position determination device according to one aspect of the present invention, the gripping position calculation unit may calculate a state of each joint of the fingers to be gripped when the object is gripped at the gripping position.
[0017] (6) Furthermore, in the gripping position determination device according to one aspect of the present invention, the gripping position calculation unit may be configured to use the initial gripping position as a starting point and search for a most stably grippable position among discrete positions between the starting point and an end point that is a predetermined position in the grippable area.
[0018] (7) In addition, in the gripping position determination device according to one aspect of the present invention, the gripping position calculation unit may be configured to use the initial gripping position as a starting point and search for a most stably grippable position for each position that is a predetermined distance away from the starting point.
[0019] (8) In order to achieve the above object, a gripping position determination system according to one embodiment of the present invention includes the gripping position determination device described in any one of (1) to (7) above, a robot hand having multiple fingers with multiple joints, an environmental sensor installed on a robot having the robot hand or in the surrounding environment of the robot and detecting a robot environmental sensor value, and a control unit that controls the movement of the robot hand to grip the object based on the gripping position determined by the gripping position determination device.
[0020] (9) In order to achieve the above-mentioned object, a grasping position determination method according to one embodiment of the present invention is a grasping position determination method for a robot hand having multiple fingers with multiple joints, in which a friction force distribution calculation unit estimates a friction force between one of the fingers to be grasped and an object from predictive control of the grasping force when the object is grasped with at least two of the fingers, and calculates a friction force distribution at which the object can be grasped on the object surface based on a value related to the friction force calculated using the estimated friction force, a graspable area selection unit selects at least one graspable area from the friction force distribution, and a grasp position calculation unit calculates a grasp position at which the object can be stably grasped from the selected graspable area.
[0021] (10) In order to achieve the above object, a program according to one embodiment of the present invention causes a computer of a grasping position determination device for a robot hand having multiple jointed fingers to estimate a frictional force between one of the fingers to be grasped and an object from predictive control of a grasping force when the object is grasped with at least two of the fingers, calculate a frictional force distribution on the object surface at which the object can be grasped based on a value related to the frictional force calculated using the estimated frictional force, select at least one graspable area from the frictional force distribution, and calculate a grasping position at which the object can be stably grasped from the selected graspable area. Effect of the Invention
[0022] According to (1) to (10), a gripping point can be determined taking into account the operation. In addition, according to (1) to (10), a robust gripping position can be determined even if there are errors in the shape or control. In addition, according to (3), the positions where slipperiness is reduced can be determined more specifically. In addition, according to (5), the robot hand can grasp the object with high precision at a grasping point that takes into account the operation. In addition, according to (6) and (7), the computational cost of determining the grasping point taking into account the operation can be reduced. [Brief description of the drawings]
[0023] [Figure 1] 1A to 1C are diagrams illustrating an example in which a gripping unit of a robot according to an embodiment grips a target object with fingers. [Diagram 2] 1 is a block diagram showing an example of the configuration of a gripping position determining system according to a first embodiment. [Diagram 3] 6A to 6C are diagrams for explaining a grip portion position determination process according to the first embodiment. [Figure 4] 4 is a flowchart of a process for determining a gripping position according to the first embodiment. [Diagram 5] FIG. 11 is a diagram for explaining a method of calculating an L2 norm in the embodiment. [Figure 6] 1A to 1C are diagrams illustrating an example of an initial gripping position, a frictional force distribution, and a corrected position when gripping a smartphone having a curved edge. [Figure 7] FIG. 13 is an image diagram of a search from an initial gripping position to a correction position. [Figure 8] FIG. 1 is a diagram showing an example of a trajectory and waypoints. [Figure 9] FIG. 11 is a block diagram showing an example of the configuration of a gripping position determining system according to a second embodiment. [Figure 10] 13 is a diagram for explaining an example of gripping position determination in the second embodiment. FIG. [Figure 11] FIG. 11 is a diagram illustrating an example of friction force distribution in the second embodiment. [Figure 12]13 is a flowchart of a process for determining a gripping position according to the second embodiment. [Figure 13] FIG. 1 is a diagram for explaining an example of a robot gripping an object with two fingers in the prior art. [Figure 14] FIG. 1 is a diagram for explaining an example of a robot gripping an object with three fingers in the prior art. [Figure 15] FIG. 13 is a diagram for explaining an example of an index ε when a robot grasps an object with two fingers with an inter-finger gap of 0.0 (m) in the prior art. [Figure 16] FIG. 13 is a diagram for explaining an example of an index ε when a robot grasps an object with two fingers with an inter-finger gap of 0.0015 (m) in the prior art. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0024] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In the drawings used in the following description, the scale of each component is appropriately changed so that each component is of a recognizable size.
[0025] [overview] FIG. 1 is a diagram showing an example in which a gripping unit 28 of a robot according to the embodiment grips a target object Obj with fingers 281. As shown in FIG. The robot includes a gripping unit 28 having at least two or more fingers 281. In the embodiment, when the gripping unit 28 of the robot grips an object Obj, the magnitude of the friction force at the gripping position is directly incorporated into an evaluation formula, and an evaluation is performed taking into account an error in the gripping position during actual gripping, thereby determining a robust gripping position even if there is an error in the shape or control. The robot also selects a graspable area based on friction and gravity, and estimates a stable graspable position within that graspable area and the posture of the grasping part (the posture of the fingers and hand joints).
[0026] figure 2 FIG. 1 is a block diagram showing an example of the configuration of a gripping position determination system according to an embodiment of the present invention. 2As shown, the gripping position determination system 1 includes a robot 2 and an environmental sensor 3. The robot 2 includes a gripping position determining device 21, a control unit 23, a memory unit 24, and a hand 25 (robot hand). The robot 2 may include, for example, legs, a head, a torso, a waist, etc. in addition to the hand 25. The robot 2 also includes a power source (not shown). The power source supplies power to each part of the robot 2. The power source may include, for example, a rechargeable battery or a charging circuit.
[0027] The gripping position determining device 21 includes an information acquiring unit 211 , an object estimating unit 212 , a frictional force distribution calculating unit 213 , a grippable area selecting unit 214 , and a gripping position calculating unit 215 . The hand 25 includes a sensor 26, a drive unit 27, and a gripper unit 28. The grip portion 28 includes a finger portion 281. The finger portion 281 includes at least two fingers. In the example of FIG. 1, the finger portion 281 includes five fingers (thumb 281a, index finger 281b, and index finger 281c). 1 b, middle finger 28 1 c, ring finger 28 1 d, little finger 28 1 The hand 25 of this embodiment is equipped with multiple jointed fingers.
[0028] The environmental sensor 3 includes an image capturing device 31, a sensor 32, an object position detection unit 33, and a communication unit .
[0029] The robot 2 and the environmental sensor 3 are connected to each other via, for example, a wireless or wired network. The robot 2 and the environmental sensor 3 may be connected to each other directly without a network.
[0030] [Functions of the gripper control system] The environmental sensor 3 is installed in a position where it can capture and detect, for example, an image of an object to be operated. The environmental sensor 3 may be provided in the robot 2 or may be attached to the robot 2. Alternatively, there may be multiple environmental sensors 3, and the environmental sensors 3 may be installed in the work environment and also attached to the robot 2. The environmental sensor 3 detects position information of the object based on the captured image and the detection result detected by the sensor, and transmits the detected object position information (environmental sensor value) to the robot 2.
[0031] The image capturing device 31 is, for example, an RGB camera. The image capturing device 31 outputs a captured image to the object position detection unit 33. In the environmental sensor 3, the positional relationship between the image capturing device 31 and the sensor 32 is known.
[0032] The sensor 32 is, for example, a depth sensor. The sensor 72 outputs a detection result to the object position detection unit 33.
[0033] The object position detection unit 33 detects the three-dimensional position, size, shape, etc. of the target object in the captured image by a known method based on the captured image and the detection result detected by the sensor. The object position detection unit 33 performs image processing (edge detection, binarization processing, feature extraction, image enhancement processing, image extraction, pattern matching processing, etc.) on the image captured by the image capture device 31 with reference to a pattern matching model stored in the object position detection unit 33 to estimate the position of the object. When multiple objects are detected from the captured image, the object position detection unit 33 detects the position of each object. The object position detection unit 33 transmits the detected object position information (environmental sensor value) to the robot remote operation control device 30 via the communication unit 34.
[0034] The communication unit 34 transmits the object position information to the robot 2.
[0035] The gripping of the robot 2 is controlled in accordance with the gripping position information determined by the gripping position determining device 21.
[0036] The control unit 23 controls the drive unit 27 based on the grip position information output by the grip position determining device 21 .
[0037] The storage unit 24 stores, for example, programs and thresholds used by the control unit 23 for control.
[0038] The sensor 26 is, for example, a joint encoder, etc. The sensor 26 is attached to, for example, each joint of the robot 2. The sensor 26 outputs the detection result to the gripping position determining device 21 and the control unit 23.
[0039] The driving unit 27 drives the gripping unit 28 of the robot 2 under the control of the control unit 23. The driving unit 27 includes, for example, an actuator, a gear, an artificial muscle, and the like.
[0040] The information acquisition unit 211 acquires object position information (environmental sensor value) from the environmental sensor 3. The information acquisition unit 211 outputs the acquired object position information to the object estimation unit 212 and the control unit .
[0041] The object estimation unit 212 estimates the position, size, shape, inclination, etc. of the grasped object by a well-known method using the object position information. Note that the object estimation unit 212 may also estimate the type and name of the object by a pattern matching method or the like.
[0042] The frictional force distribution calculation unit 213 calculates the frictional force distribution of at least one finger based on the external force (including gravity) acting on the object, the position of the contact point with a representative part of the finger (for example, the tip of the finger, the pad of the finger) on the object Obj in the object coordinate system, and the orientation (including the number) of each position of the contact point on the object Obj in the object coordinate system. In other words, the frictional force distribution calculation unit 213 calculates the force to be exerted at each contact point within the range of the control target value required for the robot 2 to grasp the object Obj against the external force. The frictional force distribution calculation unit 213 estimates the lateral force (frictional force) within a predetermined range from the initial grasping position, and calculates a value related to the frictional force using the estimated lateral force (frictional force), thereby calculating the frictional force distribution on the object surface. In other words, the frictional force distribution calculation unit 213 estimates the frictional force between the object and the finger from the predictive control of the grasping force.
[0043] The value of the friction force is, for example, the L2 norm (=√(Fx 2 +Fy 2 )). Alternatively, the value related to the friction force may be, for example, the sum of squares of the estimated friction force (Fx 2 +Fy 2 When the estimated friction force is small, take the square root Tora In some cases, the changes are greater and easier to evaluate if there is no change. Alternatively, the value of the friction force can be calculated by, for example, dividing the sum of the squares of the estimated friction force by the normal force ([Fx 2 +Fy 2 ] / Fz 2 ) The frictional force changes depending on Fz. For example, if Fz is large, the sum of the squares of the frictional force tends to be large. Even with values related to frictional force calculated in this way, there are cases where the changes are large and it is easy to evaluate. In the following embodiment, the case where the L2 norm is used as an example of a value related to frictional force will be described, but the present invention is not limited to this.
[0044] The graspable area selection unit 214 selects at least one graspable area within the area of the frictional force distribution based on the L2 norm. For example, the graspable area selection unit 214 selects an area within the area of the frictional force distribution where the L2 norm is equal to or greater than a predetermined value as the graspable area.
[0045] The grip position calculation unit 215 searches the grippable area and calculates the grip position (finger contact point) that can best resist external forces. 。
[0046] Next, a gripping portion position determination process of this embodiment will be described. In this embodiment, before gripping an object, the gripping position determination device 21 determines a gripping position of the finger portions 281 of the gripping portion 28. FIG. 3 is a diagram for explaining the grip position determination process according to the present embodiment. In FIG. 3, three fingers (thumb 281a, index finger 28 1 b, middle finger 28 1 c) is an example of grasping an object Obj. 1 b and middle finger 28 1 c) When the gripping positions are not changed and the thumb 281a is moved, the frictional force increases. Increase In other words, this is an example in which a gripping position where the grip state is stable is estimated and determined.
[0047] The arrow g11 indicates the force acting on the surface of the object Obj by the index finger 281b. The arrow g12 indicates the force acting on the surface of the object Obj by the middle finger 281c. 42 teeth ,parent This is the force acting on the surface of the object Obj by the finger 281a. The arrow g13 is the z-axis component of the thumb 281a at the initial position. The arrow g22 is the x-axis component of the thumb 281a at the initial position. The arrow g23 is the y-axis component of the thumb 281a at the initial position. Note that xyz are coordinates in the contact coordinate system.
[0048] In FIG. region g3 2 is L2 norm Shows areas related toThe darker areas indicate lower values, and the white areas indicate higher values. Such a distribution of the L2 norm is the friction force distribution g31. The area surrounded by the dashed line g51 indicates the graspable area.
[0049] In FIG. 3, when an object Obj is grasped, the grasp position of the thumb 281a where the grasp state is stable is not the initial position but the corrected position g4 where the L2 norm is the largest. 1 The arrow g42 indicates the correction position g41 is the force acting on the surface of the object Obj by the thumb 281a.
[0050] In addition, when grasping an object with opposing fingers while maintaining the object's posture, The gripping position calculated by the gripping position calculation unit 215 is the position where the friction force between the object and the finger is the smallest. It becomes. The reason will be explained with reference to FIG. 15 and FIG. 16 used in the explanation of the prior art. When grasping a coin-shaped object with two fingers, if the two fingers are opposed to each other through the object as in Figure 15, it is possible to grasp the object without using much friction force. On the other hand, if the two fingers are not opposed to each other through the object and there is a gap between the fingers as in Figure 16, it is necessary to use more friction force than in Figure 15 to grasp the object.
[0051] The graphs g923 and g924 in Figure 15 and the graphs g933 and g934 in Figure 16 show the limits of the size of the V. Also, the lines between the lines on both sides of the V (g925, g926, g935, and g936) show the force the robot can exert. In the graphs g923 and g924 in Figure 15, the lines g925 and g926, which represent the amount of force that can be exerted, are located almost halfway between the two lines of the V.
[0052] In contrast, in graph g933 of Figure 16, line g935, which represents the force that can be exerted, almost overlaps with the line on the right side of the V, indicating that the force that can be exerted is at its limit. Also, in graph g934, line g936, which represents the force that can be exerted, is close to the line on the right side of the V, indicating that the force that can be exerted is at its limit. Furthermore, the angle difference between the lines representing the allowable force (g925, g926, g935, g936) and the center of the V corresponds to the lateral force. In this way, when an object is grasped with opposing fingers while maintaining its orientation, the position at which the frictional force is minimum is the grasping position.
[0053] [Example of grip position determination processing procedure] Next, an example of a procedure for determining a gripping position will be described with reference to Fig. 4, which is a flowchart of a procedure for determining a gripping position according to this embodiment.
[0054] (Step S1) The information acquisition unit 211 acquires object position information (environmental sensor value) from the environmental sensor 3.
[0055] (Step S2) The object estimation unit 212 estimates the position, size, shape, inclination, etc. of the grasped object using the object position information by a well-known method.
[0056] (Step S3) Based on the estimated information about the object, the frictional force distribution calculation unit 213 predicts contact points between the object and each finger when the object is grasped.
[0057] (Step S4) The frictional force distribution calculation unit 213 calculates a force that may be used for gripping, based on the specifications of the robot 2 and information related to the estimated object.
[0058] (Step S5) The friction force distribution calculation unit 213 calculates a balance equation G c f c =-F e The lateral force for each candidate gripping position is obtained in the solving process.
[0059] (Step S6) The frictional force distribution calculation unit 213 calculates the L2 norm, which is the square root of the obtained lateral force, to calculate the frictional force distribution on the object surface.
[0060] (Step S7) The graspable area selection unit 214 selects, for example, an area in which the L2 norm is equal to or less than a predetermined value within the area of frictional force distribution as the graspable area.
[0061] (Step S8) The grip position calculation unit 215 searches for a grippable area and calculates information such as a grip position that can best resist an external force and the angles of the finger joints.
[0062] (Step S9) The control unit 23 generates a control command based on information such as the calculated grip position and the angles of the finger joints.
[0063] (Step S10) The control unit 23 controls the driving unit 27 to drive the gripping unit 28 of the robot 2 based on the generated control command.
[0064] The processing procedure described with reference to FIG. 4 is merely an example, and the present invention is not limited to this.
[0065] [How to calculate L2 norm] Here, the method of calculating the L2 norm will be further explained. 5 is a diagram for explaining a method of calculating the L2 norm in the embodiment. The example in Fig. 5 is an example in which a spherical object is grasped with three fingers. The frictional force distribution calculation unit 213 first calculates the force Gcfc that may be exerted during gripping. Next, the frictional force distribution calculation unit 213 solves the balance equation Gcfc=-Fe for one finger within a predetermined range from the contact point, and obtains the lateral force for each contact point obtained in the solving process. Note that Fe is an external force. The lateral force is a force parallel to the surface of the object Obj, and is a force in the lateral direction (x-axis, y-axis) in the contact point coordinate system. The lateral force thus obtained Square root of the sum of the squares of (friction force) is the L2 norm.
[0066] (Modification) Here, an example of searching for a gripping position to be corrected (correction position) using information on the shape of an object will be described with reference to Figs. 6 to 8. 6 is a diagram showing an example of an initial gripping position, a frictional force distribution, and a correction position when gripping a smartphone having a curved edge. In the example of FIG. 6, the robot 2 grips the smartphone with the thumb 281a, the index finger 28 1 b, middle finger 28 1 In this example, the gripping position of the thumb 281a is corrected so that the gripping position can be more stable when the thumb 281a is gripping the object with the index finger 28c. 1 The force acting on the surface of the object Obj by b. Arrow g102 points to the middle finger 28 1 The arrow g103 indicates the force acting on the surface of the object Obj by c. 1 This is the force acting on the surface of object Obj by a.
[0067] In this case, the gripping force distribution g1 11 For example, the initial position of the thumb is 28 1 The grip position calculation unit 215 calculates the grippable area g11 as the grippable area g11. 2 A correction position g121 based on the L2 norm is searched for within the region.
[0068] 7 is an image diagram of a search from an initial gripping position to a corrected position. In this embodiment, instead of actually gripping an object and moving a finger on the surface of the gripped object as in the conventional technology, a search is performed from the initial position g151 by calculation in a direction that provides a better cost (smaller L2 norm) (search direction g153). 61 is , which is a trajectory when a finger is moved according to a search. Point g161 is a waypoint. The trajectory is, for example, a path that the tip of a finger should take from the initial position (start point g151) to the end point g152. The grip position calculation unit 215 solves the equation of motion (balance equation) of the fingertip contact point at the waypoint of such a search path (intermediate position) to calculate an optimal grip position.
[0069] The search route on which the waypoints are set is a route that detours around positions on the object that the finger must avoid (for example, positions with protrusions). The waypoints and routes may be set sequentially, for example, by moving the candidate gripping positions by a predetermined amount. In this case, the initial position to be searched is determined, but the end point is not. Also, although it is preferable that the waypoints are spaced at equal intervals, they do not have to be spaced at equal intervals. According to this embodiment, by performing such a search, it is possible to reduce the calculation cost of determining a gripping point taking into account an operation.
[0070] Fig. 8 is a diagram showing an example of a trajectory and waypoints. Fig. 8 shows an example of grasping the surface of an object with the tip of a finger. In Fig. 8, a model g201 represents a model of a hand 25 including one finger. The finger is modeled with a first joint g202 that connects a proximal segment g203 to the palm, and a second joint g204 that connects a distal segment g205 to the proximal segment g203.
[0071] During the search, the grip position calculation unit 215 determines discrete waypoints (e.g., point g213) on the trajectory that the tip of the finger should take from the initial position (start point g211) to the end point g212. As shown in FIG. 7, the angle of the finger joint (and further the angle of the joint of the hand 25) changes depending on where the finger touches (tip, pad, etc.). For this reason, the grip position calculation unit 215 searches for the grip position and also determines the angles of the joints of the hand 25 and the finger joints, and outputs the determined results to the control unit 23. The number of waypoints is, for example, 100, and may be less or more than this. In this way, the trajectory that the tip of the fingertip should take to move from the initial contact point to the correction position for correcting the grip position can be narrowed down to within the grippable area.
[0072] In addition, in this embodiment, discrete key points are defined on the trajectory, and a predetermined solution space is searched for at the waypoints using the equation of motion for the fingertip gripping position (the equation of motion for the contact between a curved surface extending from the tip to the pad of the fingertip and the surface of an object), and it is evaluated whether the sum of the frictional forces of all fingers is within an acceptable range when the fingertip contact point (for example, the tip or pad) is changed, and control is possible while modifying the contact point of the fingertip tip that is controllable within a reasonable range. In this embodiment, the evaluation index ε is added to this equation of motion, and a graspable area is found from the friction force distribution based on the result of the obtained balance equation, and the optimal position within this graspable area is found.
[0073] As described above, in this embodiment, an estimated friction force is obtained from predictive control of the grip force, and a friction force distribution of a specific finger is calculated on the object surface from the L2 norm of the friction forces of all the fingers obtained. In this embodiment, at least one or more grippable areas are determined from the friction force distribution, and a contact point that can most resist an external force is calculated from the grippable areas. In this embodiment, the friction force is not actually measured by touching, but a grip position in a state where the object can be stably held is determined.
[0074] As a result, according to this embodiment, when manipulating an object, if the fingertip shape is curved, the grasping position can be evaluated based on the entire trajectory as it rolls along the curved surface, making it possible to determine the grasping point taking the operation into consideration.
[0075] In the conventional technique described in Patent Document 2, when the surface of an object is curved, it is not possible to obtain an appropriate frictional force even if a finger is moved in contact with the object. In contrast, in this embodiment, instead of obtaining the frictional force while actually contacting and moving a finger with the object, the frictional force in the gripping state is predicted and the contact point that can most resist the external force is calculated, so that a gripping position that allows stable gripping can be appropriately determined.
[0076] <Second embodiment> In this embodiment, a gripping position that reflects a specific condition (e.g., gravity) is calculated to maintain the posture and movement of the object in a specific state. In this case, the gripping position is not necessarily the position where the frictional force is minimum. In this embodiment, it also includes a state where the object is balanced with respect to gravity and can be gripped, and a state where the object can be gripped even if the posture of the object is changed, such as tilted.
[0077] 9 is a block diagram showing an example of the configuration of a gripping position determining system according to this embodiment. As shown in FIG. 9, the gripping position determining system 1A includes a robot 2A and an environment sensor 3. The robot 2A includes a gripping position determining device 21A, a control unit 23, a memory unit 24, and a hand 25. The gripping position determining device 21A includes an information acquiring unit 211, an object estimating unit 212, a frictional force distribution calculating unit 213A, a grippable area selecting unit 214A, and a gripping position calculating unit 215.
[0078] The frictional force distribution calculation unit 213A determines whether or not there is a graspable region that can be grasped in a specific state, based on the constraint conditions.
[0079] The graspable region selection unit 214A selects at least one graspable region within the region of the frictional force distribution based on the constraint conditions and the L2 norm. The constraint conditions include, for example, the force that the hand 25 may exert, the minimum force that each finger should exert equally, whether or not the posture should be maintained, etc.
[0080] Fig. 10 is a diagram for explaining an example of gripping position determination in this embodiment. In Fig. 10, the object Obj is a cylindrical object. In the example of Fig. 10, the object is gripped from above with four fingers (thumb, index finger, middle finger, and ring finger).
[0081] Position G301, index finger 28 1 b is the gripping position of the middle finger 28. 1 The position g303 is the gripping position of the ring finger 28 c. 1 Position 304 is the gripping position of thumb 28 d. 1The initial gripping position of a is shown in Fig. 1. Each arrow indicates the force acting on the object Obj when it is gripped by each finger.
[0082] Here, in order to keep the gravitational balance of the object Obj intact, that is, to hold the object without tilting, the thumb 28 1 The initial grip position of a is 28 mm for the index finger. 1 b gripping position and middle finger 28 1 c Grasping position and ring finger 28 1 In this case, the frictional force distribution calculation unit 213 calculates the gripping force distribution g311 in a predetermined range in the xy directions of the object Obj, i.e., in the circumferential direction. Then, the gripping position determination device 21A calculates the corrected gripping position g312 as a position where the object can be stably held.
[0083] Next, when the object Obj is to be held so that it is okay for the gravitational balance of the object Obj to be disturbed, i.e., the object is allowed to tilt, the gripping position determination device 21A searches (search path g321) for a position on the surface of the object Obj where the object can be gripped stably, including a predetermined range in the z-axis direction of the object Obj, and where the L2 norm is minimum, and calculates the gripping position g322.
[0084] [Example of friction force distribution] Next, an example of the friction force distribution in this embodiment will be described. Fig. 11 is a diagram showing an example of frictional force distribution in this embodiment. In Fig. 11, the object Obj is a spherical object. In the example of Fig. 11, the object is held from the side with two fingers (thumb and index finger).
[0085] Position G401, index finger 28 1 b. Position 402 is the gripping position of thumb 28 1 The initial gripping position of a is shown in Fig. 1. Each arrow indicates the force acting on the object Obj when it is gripped by each finger.
[0086] In this case, the frictional force distribution g411 is distributed in the z-axis direction in addition to the xy directions of the object Obj. The first graspable region g421 is a region based on the L2 norm. In the first graspable region g421, a gripping position that allows stable gripping is position g422. This position g422 is a position facing the index finger, and is selected, for example, when the effect of gravity is weakened.
[0087] The second graspable region g431 is a region when the case where the object can be grasped stably in the gravitational force is also considered as a constraint. In this second graspable region g431, the gripping position where the object can be grasped stably is position g432. Position g432 is a position that is stable in the gravitational direction even if it is slightly lower than position g422, and is selected, for example, when a change in the posture of the object is allowed.
[0088] As indicated by the arrow g441, the range of the graspable area indicates the area that can be grasped. In this embodiment, at least one graspable area is determined from the frictional force distribution, and a contact point that can most resist gravity and external forces is calculated from among the graspable areas. Then, in this embodiment, a position where the frictional force generated by all fingers in the lateral direction of the object is minimized is searched for and corrected by moving the initial gripping position of one finger from among at least one graspable area.
[0089] [Example of grip position determination processing procedure] Next, an example of a procedure for determining a gripping position will be described with reference to Fig. 12, which is a flowchart showing the procedure for determining a gripping position according to this embodiment.
[0090] (Steps S1 to S5) The gripping position determining device 21A performs the processes of steps S1 to S5.
[0091] (Step S101) The frictional force distribution calculation unit 213 calculates the L2 norm, which is the square root of the obtained lateral force, to calculate the frictional force distribution on the object surface.
[0092] (Steps S7 to S8) The gripping position determining device 21A performs the processes of steps S7 to S8.
[0093] (Step S102) The frictional force distribution calculation unit 213A determines whether or not there is a graspable area that can be grasped in a specific state based on the constraint conditions. If the frictional force distribution calculation unit 213A determines that there is a graspable area that can be grasped in a specific state (Step S102; YES), the frictional force distribution calculation unit 213A proceeds to the process of Step S103. If the frictional force distribution calculation unit 213A determines that there is no graspable area that can be grasped in a specific state (Step S102; NO), the frictional force distribution calculation unit 213A proceeds to the process of Step S104.
[0094] (Step S103) The gripping position calculation unit 215 searches for a grippable area and calculates other gripping positions that can best resist an external force and information such as the angles of the finger joints.
[0095] (Steps S9 to S10) The control unit 23 performs the processes of steps S9 to S10.
[0096] The frictional force distribution calculation unit 213A performs the processes of steps S102 to S103 for the graspable areas that can be grasped in all specific states.
[0097] In this manner, in this embodiment, when there are multiple gripping methods that balance the external force and gravity, one of the multiple grippable areas is selected at a time, and a gripping position is calculated for each selected grippable area. The gripping position calculation unit 215 selects one of the multiple gripping positions calculated in this manner, depending on the state of the object, constraint conditions, and the like.
[0098] This makes it possible to determine multiple grip positions that allow stable gripping. As a result, according to this embodiment, evaluation that takes into account grip position errors can be considered as an expected value that takes into account probability distribution, and by treating shape elements on the fingertip side and crushing of the epidermis as distributions, it becomes possible to perform evaluation while taking into account multiple elements without tuning the weighting parameters.
[0099] In addition, for the two or more gripping positions described above, when the posture of the object changes (for example, when carrying it), the gripping position selected from the multiple calculated gripping positions may be switched depending on the posture of the object.
[0100] When an object is operated, if the fingertip shape is curved, the grip point moves while rolling. As described above, according to the first and second embodiments, the entire trajectory is evaluated, making it possible to determine the grip point while considering the operation. Furthermore, since the evaluation considering the error in the grip position can be considered as an expected value considering a probability distribution, as described above, according to the first and second embodiments, by treating the shape elements on the fingertip side and the crushing of the epidermis as a distribution, it becomes possible to perform evaluation while considering multiple elements without tuning the weighting parameters.
[0101] Furthermore, according to the first and second embodiments, unlike the evaluation of gripping performance using all fingers in the conventional technology, the frictional force information of all fingers is incorporated into the evaluation of the quality of the contact point of one finger, and the contact point can be determined from the distribution on the object surface. As a result, according to the first and second embodiments, it has become possible to more specifically determine the position where it is less likely to slip. Furthermore, according to the first and second embodiments, an evaluation is performed that takes into account errors in the gripping position during actual grasping in a direction that makes it easier to remain inside the graspable ellipsoid, making it possible to determine a robust gripping position even if there are errors in shape or control.
[0102] In addition, a program for realizing all or part of the functions of the gripping position determination device 21 (or 21A) in the present invention may be recorded in a computer-readable recording medium, and the program recorded in the recording medium may be read into a computer system and executed to perform all or part of the processing performed by the gripping position determination device 21 (or 21A). Note that the term "computer system" as used herein includes hardware such as an OS and peripheral devices. The term "computer system" also includes a system built on a local network and a system built on the cloud. The term "computer-readable recording medium" refers to portable media such as a flexible disk, a magneto-optical disk, a ROM, a CD-ROM, and a storage device such as a hard disk built into a computer system. The term "computer-readable recording medium" also includes a storage device that holds a program for a certain period of time, such as a volatile memory (RAM) inside a computer system that becomes a server or a client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line.
[0103] The above program may also be transmitted from a computer system in which the program is stored in a storage device or the like to another computer system via a transmission medium, or by transmission waves in the transmission medium. Here, the "transmission medium" that transmits 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. The above program may also be for realizing part of the above-mentioned functions. Furthermore, it may be a so-called difference file (difference program) that can realize the above-mentioned functions in combination with a program already recorded in the computer system.
[0104] The above describes the form for carrying out the present invention using an embodiment, but the present invention is not limited to such an embodiment, and various modifications and substitutions can be made within the scope that does not deviate from the gist of the present invention. [Explanation of symbols]
[0105] 1, 1A...Grasp position determination system, 2...Robot, 3...Environment sensor, 21, 21A...Grasp position determination device, 23...Control unit, 24...Memory unit, 25...Hand, 211...Information acquisition unit, 212...Object estimation unit, 213, 213A...Friction force distribution calculation unit, 214, 214A...Graspable area selection unit, 215...Grasp position calculation unit, 26...Sensor, 27...Drive unit, 28...Grasp unit, 281...Finger unit, 281a...Thumb, 28 1 b…index finger, 28 1 c...middle finger, 28 1 d…Ring finger, 28 1 e...little finger, 31...photographing device, 32...sensor, 33...object position detection unit, 34...communication unit
Claims
1. In a gripping position determination device for a robot hand having multiple joints and fingers, a frictional force distribution calculation unit that calculates a frictional force distribution that allows one of the fingers to grasp an object when the object is grasped with at least two of the multiple fingers, based on a predetermined external force acting on the object, positions of contact points with representative parts of the fingers on the object in an object coordinate system, and orientations of each position of the contact points on the object in the object coordinate system; and a graspable area selection unit that selects at least one graspable area from the frictional force distribution; a gripping position calculation unit that calculates a contact point of the finger at which the object can be stably gripped from the selected grippable area as a gripping position, the frictional force distribution calculation unit estimates a lateral force, which is a frictional force within a predetermined range, from an initial gripping position at which one of the fingers grips the object, and calculates a value related to the frictional force using the estimated lateral force to calculate a frictional force distribution on a surface of the object; The value related to the friction force includes an L2 norm, which is the square root of the sum of the squares of the lateral forces estimated in two axial directions along the surface of the object; The graspable area selection unit selects an area in which the L2 norm is equal to or greater than a predetermined value within the area of the friction force distribution as a graspable area, The gripping position calculation unit calculates, as the gripping position, a finger contact point that can most resist the external force within the grippable area. Grip position determination device.
2. The gripping position calculation unit selects the grippable area in which a posture or movement of the object can be maintained in a specific state, and calculates a gripping position within the grippable area in which the posture or movement of the selected object can be maintained in a specific state. The grip position determining device according to claim 1 .
3. the gripping position calculation unit calculates a state of each joint of the fingers to be gripped when the object is gripped at the gripping position. The gripping position determining device according to claim 1 or 2.
4. The gripping position calculation unit sets the initial gripping position as a starting point, and searches for a most stably grippable position at discrete positions between the starting point and an end point up to a predetermined position in the grippable area. The grip position determining device according to claim 1 .
5. The gripping position calculation unit sets the initial gripping position as a starting point and searches for a most stably grippable position for each position at a predetermined interval from the starting point. The grip position determining device according to claim 1 .
6. The gripping position determining device according to any one of claims 1 to 5, A robot hand having multiple joints and multiple fingers; an environmental sensor that is installed in a robot having the robot hand or in a surrounding environment of the robot and detects a robot environmental sensor value; a control unit that controls a movement of the robot hand to grip the object based on the grip position determined by the grip position determination device; A gripping position determination system comprising:
7. A method for determining a gripping position of a robot hand having multiple joints and fingers, comprising: a frictional force distribution calculation unit calculates a frictional force distribution capable of gripping an object with one of the fingers when the object is gripped with at least two of the multiple fingers, based on a predetermined external force acting on the object, positions of contact points with representative portions of the fingers on the object in the object coordinate system, and orientations of each position of the contact points on the object in the object coordinate system; a grippable area selection unit selecting at least one grippable area from the friction force distribution; a gripping position calculation unit calculates a contact point of the finger at which the object can be stably gripped from the selected grippable area as a gripping position; a lateral force, which is a frictional force within a predetermined range, is estimated from an initial gripping position at which one of the fingers grips the object, and a value related to the frictional force is calculated using the estimated lateral force, thereby calculating a frictional force distribution on a surface of the object; The value related to the friction force includes an L2 norm, which is the square root of the sum of the squares of the lateral forces estimated in two axial directions along the surface of the object; A region in which the L2 norm is equal to or greater than a predetermined value within the friction force distribution region is selected as a graspable region; A finger contact point that can most resist the external force is calculated as the gripping position in the grippable area. Grasp position determination method.
8. A computer of a gripping position determination device for a robot hand having multiple joints and multiple fingers, calculating a frictional force distribution that allows one of the fingers to grasp an object when the object is grasped with at least two of the multiple fingers, based on a predetermined external force acting on the object, positions of contact points with representative portions of the fingers on the object in an object coordinate system, and orientations of each position of the contact points on the object in the object coordinate system; selecting at least one graspable region from the frictional force distribution; calculating a contact point of the finger at which the object can be stably grasped from the selected graspable region as a grasp position; a lateral force, which is a frictional force within a predetermined range, is estimated from an initial gripping position at which one of the fingers grips the object, and a value relating to the frictional force is calculated using the estimated lateral force, thereby calculating a frictional force distribution on a surface of the object; The value related to the friction force includes an L2 norm, which is the square root of the sum of the squares of the lateral forces estimated in two axial directions along the surface of the object; A region in which the L2 norm is equal to or greater than a predetermined value within the region of the friction force distribution is selected as a graspable region; A finger contact point that can most resist the external force is calculated as the gripping position in the grippable area. program.
Citation Information
Patent Citations
Qualitative inference processing method
JP1989076358A
Gripping force control method of robot hand
JP2005144573A
Teaching device and teaching method
JP2007319997A
Gripping position calculating device and gripping position calculating method
JP2010179443A
