Grinding amount estimation device

Abstract

A grinding amount estimation device (50) is provided that can facilitate the setting of parameters of a teaching trajectory and force control in a grinding operation. A grinding amount estimation device (50) estimates the grinding amount in a grinding operation in which a grinding tool mounted on a robot manipulator is brought into contact with a target workpiece by force control, the grinding amount estimation device (50) comprising: a storage unit storing an action program; and a grinding amount estimation unit (56) estimating the grinding amount based on at least one of an action trajectory of the grinding tool, an action speed of the grinding tool, and a pressing force of the grinding tool on the target workpiece obtained according to the action program.

Description

Technical Field

[0001] The present invention relates to a grinding amount estimation device. Background Art

[0002] A robot manipulator is equipped with a force sensor, which can detect the force applied to a workpiece and perform force control while performing advanced operations such as exploration, embedding, and grinding. For such a robot system, there is also a known system configured to display the force detected by the force sensor (for example, refer to Patent Document 1).

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Publication No. 2017-1122 Summary of the invention

[0006] Problem that the invention aims to solve

[0007] However, in order to properly perform force control operations such as grinding operations, skilled parameter adjustment capabilities are required. Generally speaking, in order to perform such parameter adjustments, operators need to repeatedly fail / succeed in force control to master the technical secrets of parameter setting. A grinding amount estimation device that can facilitate the teaching trajectory and force control parameter setting in grinding operations is desired.

[0008] Solutions for solving problems

[0009] One embodiment of the present disclosure is a grinding amount estimation device that estimates the grinding amount in a grinding operation performed by bringing a grinding tool mounted on a robot manipulator into contact with a target workpiece through force control, the grinding amount estimation device comprising: a storage unit that stores an action program; and a grinding amount estimation unit that estimates the grinding amount based on at least one of an action trajectory of the grinding tool obtained according to the action program, an action speed of the grinding tool, and a pressing force of the grinding tool on the target workpiece.

[0010] Effects of the Invention

[0011] According to the above configuration, the operator can intuitively grasp the estimated grinding amount and can easily adjust the teaching trajectory and force control parameters.

[0012] These and other objects, features and advantages of the present invention will become more apparent from the detailed description of typical embodiments of the present invention shown in the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 11 is a system configuration diagram of a robot system including a control device as a simulation device according to one embodiment.

[0014] Figure 2 A configuration example of a robot system is shown.

[0015] Figure 3 Another configuration example of the robot system is shown.

[0016] Figure 4 This is a functional block diagram of a control device, an external computer, and a display device.

[0017] Figure 5 This is a block diagram of the force control in the robot motion control unit.

[0018] Fig. 6A This is a diagram for explaining the correlation between the robot's motion trajectory and the grinding amount.

[0019] Figure 6B This is a diagram for explaining the correlation between the robot's motion trajectory and the grinding amount.

[0020] Fig. 7A This is a diagram for explaining the correlation between the pressing force and the grinding amount.

[0021] Figure 7B This is a diagram for explaining the correlation between the pressing force and the grinding amount.

[0022] Fig. 8A This is a diagram for explaining the correlation between the operating speed and the grinding amount.

[0023] Figure 8B This is a diagram for explaining the correlation between the operating speed and the grinding amount.

[0024] Fig. 9 An example of an image representing a virtual pressing force is shown.

[0025] Fig.10 Another example of an image representing a virtual pressing force is shown.

[0026] Fig.11 This is an example of an augmented reality image in which an image showing a virtual pressing force and an estimated grinding amount is displayed in a manner superimposed on a real image.

[0027] Fig.12 This is another example of an augmented reality image in which an image showing a virtual pressing force and an estimated grinding amount is displayed in a manner superimposed on a real image.

[0028] Fig.13 It is shown that an image showing the recommended track is also superimposed on Fig.11 Figure 1 shows an example of such an image.

[0029] Fig.14 It is shown that an image indicating the recommended speed is also superimposed on Fig.11 Figure 1 shows an example of such an image.

[0030] Fig.15 This is a diagram for explaining the polishing area.

[0031] Fig.16 This is a diagram for explaining the polishing area.

[0032] Fig.17 This is a diagram for explaining the calculation method of the polishing area.

[0033] Fig.18 This is a diagram showing a first example regarding the type of grinding tool and the grinding amount.

[0034] Fig.19 This is a diagram showing a second example regarding the type of grinding tool and the grinding amount.

[0035] Fig. 20 This is a diagram showing a third example regarding the type of grinding tool and the grinding amount. DETAILED DESCRIPTION

[0036] Next, the embodiments of the present disclosure are described with reference to the accompanying drawings. In the referenced drawings, the same structural parts or functional parts are marked with the same figure marks. For easy understanding, the scales are appropriately changed in these drawings. In addition, the method shown in the drawings is an example for implementing the present invention, and the present invention is not limited to the method shown in the drawings.

[0037] Figure 1 FIG. 1 is a system configuration diagram of a robot system 100 including a control device 50 of a grinding amount estimation device according to one embodiment. Figure 1 As shown, the control device 50 is connected to a robot manipulator 10 (hereinafter referred to as the manipulator 10) equipped with a tool at the front end of the wrist and a force sensor 3 as a force detector for detecting the external force applied to the tool. The force sensor 3 is installed between the front end of the wrist of the manipulator 10 and the tool. The control device 50 has a force control function, which enables the manipulator 10 to perform various operations such as exploration operations, precision embedding operations, and grinding as advanced operations while detecting the force applied to the workpiece. In addition, the control device 50 may also have a structure as a general computer, which has a CPU, ROM, RAM, a storage device, an operating unit, a display unit, an input / output interface, a network interface, etc.

[0038] Furthermore, the control device 50 is connected to an external computer 90 and a display device 70. The external computer 90 has a function of performing a physical simulation based on a motion model of the manipulator 10 when the control device 50 performs a simulation of a force control operation (hereinafter referred to as a force control simulation). The display device 70 displays the result of the force control simulation. In addition, in this specification, the simulation includes not only the operation of calculating the position of the manipulator by numerical simulation, but also the operation of causing a shape model of the manipulator to operate in a simulated manner according to teaching data.

[0039] exist Figure 2 and Figure 3 An example of the structure of the robot system 100 is shown in FIG. Figure 2 and Figure 3 In FIG. 1 , only the manipulator 10 (including the force sensor 3 and the tool portion 11 ) and the target workpiece are shown. Figure 2 FIG. 1 shows a configuration example in which a grinder 8 for grinding a workpiece W1 is mounted on the tool portion 11. A disc-shaped grinding stone 9 is mounted on the grinder 8. The grinder 8 is suitable for Figure 2 As shown in FIG. 1 , the burr 81 on the upper surface of the target workpiece W1 is ground. Figure 3 The structure example of the grinder 18 having a triangular pyramid-shaped grinding stone 19 mounted on the tool part 11 is shown. The grinder 18 is suitable for Figure 3 As shown in FIG. 1 , the burr 81B formed on the side surface of the target workpiece W2 is ground.

[0040] The control device 50 has a function of estimating the grinding amount when the grinding operation is performed according to the teaching data (operation program), and displaying the estimated grinding amount result as an AR (augmented reality) image or a VR (virtual reality) image on the display device 70. Thus, for example, the operator can understand the degree of grinding amount before actually performing the grinding operation, and can adjust the teaching data, force control parameters, etc.

[0041] Figure 4 is a functional block diagram of the control device 50, the external computer 90, and the display device 70. Figure 4As shown in FIG. 1 , the control device 50 includes a storage unit 51 storing various information, a force control simulation execution unit 52 in charge of executing force control simulation, a robot motion control unit 53 controlling the motion of the robot manipulator 10, a virtual force generator (virtual force generation unit) 54, a virtual force learning unit 55, a grinding amount estimation unit 56 executing a calculation for estimating the grinding amount, a grinding amount learning unit 57, a recommended value generation unit 58, and a tool selection unit 59. The storage unit 51 stores the motion program of the robot manipulator 10, 3D model data of the manipulator 10, the tool, and the workpiece, force control parameters, and other various data used in controlling the manipulator 10. The virtual force generator 54 generates a virtual force that the tool part 11 receives from the target workpiece when the tool part 11 is in contact with the target workpiece based on the position information of the tool part 11 obtained from the motion program or the simulation result of the force control motion. In this specification, the force virtually obtained as the force acting on the object may be described as a virtual force, and in the case where it is a pressing force, it may also be described as a virtual pressing force.

[0042] The external computer 90 includes a physical simulation unit 91 that executes physical simulation of the manipulator 10 based on a motion model (motion equation) of the manipulator 10 .

[0043] The display device 70 is configured as a head-mounted display in this embodiment. In addition, the display device 70 can also be configured by other information processing devices such as a tablet terminal equipped with a camera. The operator wears the display device 70 configured as a head-mounted display. The display device 70 has a camera 71, an AR / VR image processing unit 72 that performs image processing for displaying an augmented reality (AR) image or a virtual reality (VR) image, a display 73, and a sound output unit 74. The camera 71 is arranged on the display device 70 in a manner such that the optical axis of the camera lens faces forward, and the camera 71 captures an image of a real working space including the manipulator 10. The AR / VR image processing unit 72 uses the information of the estimated grinding amount obtained by the grinding amount estimation unit 56 to perform augmented reality image processing that superimposes an image representing the estimated grinding amount on a real image, or virtual reality image processing that superimposes an image representing the estimated grinding amount on an image (motion image animation) of a virtual reality space in which a model of each object such as the manipulator 10 is configured. The display 73 is arranged in front of the wearer's eyes and displays the image (image) generated by the AR / VR image processing unit 72.

[0044] Figure 5This is a block diagram of force control in the robot motion control unit 53. In this embodiment, the direction in which "force control + position control" should be performed (the direction in which the workpiece is pressed by the tool) and the direction in which only position control should be performed are divided, and the speed (angular velocity) command calculated for the direction in which "force control + position control" should be performed is synthesized with the speed (angular velocity) command calculated for the direction in which only position control should be performed to control the manipulator 10. Figure 5 For the sake of convenience, the position control is omitted in the figure, but the position control is performed based on the position control side (for example, PD control) generally known in the art, which performs position control by feeding back the position detection value through the position sensor provided on each axis of the manipulator 10. Figure 5 In the force control shown, the difference between the target force (force + torque) in the pressing direction and the force (torque) acting on the workpiece detected by the force sensor 3 is multiplied by the control parameter of the force control gain to calculate the command speed (angular velocity). Here, the force control gain represents the performance of the force control, and has the property that the larger the value of the force control gain, the faster the position / posture is corrected. The detection of the force (torque) and the calculation of the corresponding speed (angular velocity) command amount are performed in each control cycle. The force control rule (the calculation formula of the speed (angular velocity) command amount) in this case can be expressed as follows.

[0045] Δx=Kf(F-Fd)

[0046] Where, Kf: force control gain

[0047] Fd: target force (force + torque, force: Fx, Fy, Fz, torque: Mx, My, Mz)

[0048] F: Detected force

[0049] Δx: Target movement amount (speed) per control cycle

[0050] Next, a description will be given of a method of estimating the polishing amount performed by the polishing amount estimating unit 56. The polishing amount estimating unit 56 estimates the polishing amount in the polishing operation using the polishing amount estimating method 1 or 2 described below.

[0051] (Grinding amount estimation method 1): The robot's motion trajectory, motion speed, and pressing force are considered as parameters that have a correlation with the grinding amount. In the grinding amount estimation method, one of these parameters is used to derive the correlation with the grinding amount by linear approximation or curve approximation. In addition, in this specification, when simply referred to as "motion trajectory", it includes a trajectory obtained by so-called teaching, that is, a teaching trajectory and a motion trajectory of the manipulator 10 (tool front end) obtained by numerical simulation or the like. A virtual force (virtual pressing force) generated by the method described later is used as the pressing force for grinding amount estimation.

[0052] (Grinding amount estimation method 2): Collect training data that correlates the robot's motion trajectory, motion speed, and pressing force with the grinding amount, and construct a learning model that correlates these parameters with the grinding amount through machine learning.

[0053] The grinding amount estimation method 1 will be described. First, the correlation between the movement trajectory, movement speed, and pressing force of the robot and the grinding amount will be described. Fig. 6A and Figure 6B This is a diagram for explaining the correlation between the robot's motion trajectory (here, the teaching trajectory) and the grinding amount. Fig. 6A The teaching trajectory L1 shown in FIG. 1 shows that the teaching trajectory L1 for the grinding stone 9 is an appropriate trajectory and the grinding amount is also appropriate. Figure 6B In the embodiment, the teaching track L2 of the grindstone 9 is farther from the surface of the target workpiece W1, especially near the protrusion 82. As described above, when the teaching track (teaching point) is farther from the target workpiece, the grinding amount is reduced.

[0054] Fig. 7A and Figure 7B This is a diagram for explaining the correlation between the pressing force and the grinding amount. Fig. 7A The setting of the pressing force (pressing force F71) is appropriate and the grinding amount is also appropriate. Figure 7B In the setting of pressing force (pressing force F72) and Fig. 7A In this case, even if the teaching track and teaching speed are the same, the grinding amount is reduced.

[0055] Fig. 8A and Figure 8B This is a diagram for explaining the correlation between the operating speed of the tool (the moving speed of the grindstone along the teaching trajectory) and the grinding amount. Fig. 8A This indicates that the setting of the action speed (teaching speed) is appropriate and the grinding amount is also appropriate. Figure 8B In the setting of action speed and Fig. 8A The situation becomes faster than that of Figure 8B In the case of Fig. 8A The time required for grinding is reduced compared to the case where the grinding is performed with a smaller grinding wheel, so even if the teaching trajectory and pressing force are the same, the grinding amount is reduced.

[0056] As described above, the robot's motion trajectory, motion speed, and pressing force are respectively correlated with the grinding amount. Therefore, the grinding amount can be estimated using any one of a calculation model that linearly approximates or curves the correlation between the robot's motion trajectory (the distance between the motion trajectory and the surface of the target workpiece) and the grinding amount (polynomial approximation of quadratic or higher, logarithmic approximation, etc.) based on measured data, a calculation model that linearly approximates or curves the correlation between the robot's motion speed and the grinding amount based on measured data, and a calculation model that linearly approximates or curves the correlation between the pressing force and the grinding amount based on measured data. In addition, such linear approximation or curve approximation of the correlation can also be performed according to the type of the target workpiece and the type of abrasive (grindstone). The correlation between two or more variables of the robot's motion trajectory, motion speed, and pressing force and the grinding amount can also be predicted by multivariate regression analysis.

[0057] In the grinding amount estimation method 1, the virtual pressing force acting on the target workpiece during the grinding operation is obtained based on the positional relationship between the teaching trajectory and the target workpiece, or by virtual force generation methods 1 to 3 described below.

[0058] (Virtual force generation method 1): Set the motion model (motion equation) of the robot manipulator 10 and execute it through physical simulation Figure 5 The action of the force control block diagram shown in FIG. Based on the position of the tool tip obtained by physical simulation, the virtual pressing force acting on the target workpiece is obtained by using a calculation model. That is, in the case of the virtual force generation method 1, the structure is as follows: Figure 5 As shown, a motion model is set for the manipulator 10, and a virtual pressing force is calculated by the virtual force generator 54. That is, the virtual force generator 54 functions as a force sensor in the force control simulation.

[0059] (Virtual force generation method 2): A virtual force (virtual pressing force) is obtained by using log data including force (torque) detected by the force sensor 3 when a force control-based operation is performed in the same action environment as in the past, position information of the robot (manipulator 10), or log data obtained by detecting and recording the force (torque) acting on the workpiece by the force sensor while stopping the drive of the tool (e.g., the rotation drive of a grinding stone) while the robot is actually moved relative to the target workpiece using an action program. In the case of virtual force generation method 2, the distance between the tool and the target workpiece can be obtained based on the teaching trajectory. If there is log data in the log data that the distance between the robot's action trajectory and the target workpiece is the same, the pressing force recorded as the log data is used as the virtual force (virtual pressing force).

[0060] (Virtual force generation method 3): In actual operations related to a specific workpiece, training data representing the correspondence between the relative position and speed between the robot (tool) and the workpiece and the force (torque) detected by the force sensor is collected, and a learning model is constructed through the learning function to calculate the virtual force (virtual pressing force).

[0061] The virtual force generation method 1 is described in detail. In the virtual force generation method 1, the motion equation (motion model) of the robot manipulator 10 is set, and the virtual force generation method 1 is used to generate the virtual force. Figure 5 The force control block shown is operated to obtain the position of the robot manipulator 10 (the position of the tool tip). The equation of motion of the robot manipulator 10 is generally expressed by the following equation.

[0062] [Number 1]

[0063]

[0064] In the above formula, θ represents the angle of each joint, M is the matrix related to the moment of inertia, h is the matrix related to the Coriolis force and centrifugal force, g is the term representing the influence of gravity, τ is the torque, and τ L is the load torque.

[0065] The motion instructions based on the teaching track (in Figure 5 In the example, the command given to the manipulator 10) is given as input data to the motion equation to calculate the robot's behavior (the position of the tool tip). Based on the tool tip position calculated according to the above motion equation, the virtual force (virtual pressing force) F received by the workpiece when the tool tip position contacts the target workpiece is calculated. The calculation example of the virtual force F is shown below.

[0066] The first calculation example of the virtual force (virtual pressing force) F is an example in which the rigidity of the target workpiece is relatively low relative to the tool. In this example, the amount by which the tool tip position exceeds the contact position with the target workpiece and moves toward the target workpiece can be set as δ, which is multiplied by a coefficient Kd related to the rigidity of the workpiece to be obtained by the following formula.

[0067] F=Kd·δ…(1a)

[0068] Alternatively, there can be a method as follows: when the position of the tool tip contacts the target workpiece, the force received from the workpiece is set to F, and the speed when the tool tip exceeds the contact position with the workpiece is set to Vc, and the force F received from the workpiece is calculated by the following formula.

[0069] F=Kd·δ+Kc·Vc…(1b)

[0070] These coefficients Kd and Kc can be set according to the rigidity, shape, etc. of the target workpiece.

[0071] The second calculation example of the virtual force (virtual pressing force) F is an example of calculating the virtual force F based on the deflection amount of the tool when the rigidity of the tool is relatively low relative to the target workpiece. The amount δ by which the tip of the tool moves toward the target workpiece beyond the contact position with the target workpiece is regarded as the deflection amount of the tool, and the rigidity coefficient (virtual spring coefficient) of the tool is used to calculate the virtual force F by the following formula.

[0072] F = (virtual spring coefficient of the tool) × δ…(2a)

[0073] In addition, when the tool is a so-called floating tool having a mechanism (spring mechanism) that extends and retracts in the pressing direction, the extension length of the tool tip can be calculated based on the tool tip position and the position of the target workpiece, and the virtual force F can be obtained by the following formula.

[0074] F = (spring constant of tool) × telescopic length…(2b)

[0075] The third calculation example of the virtual force (virtual pressing force) F is an example of calculating the virtual force F based on the distance moved by the robot (tool tip) in response to the speed command in the pressing direction when the rigidity of the tool is relatively high. In this example, the moving position based on the speed command is set to Tx, and the position after the robot (tool tip) actually moves in response to the speed command is set to d, and the virtual force F is calculated by the following formula.

[0076] F = k × (Tx-d)…(3)

[0077] Here, k is a coefficient. The coefficient k may be set to a value obtained as an experimental value, an empirical value, or the like.

[0078] In the above-mentioned calculation example, the virtual force may be obtained by using teaching data (teaching trajectory, teaching speed) instead of the position and speed of the tool tip obtained by physical simulation.

[0079] Next, the virtual force generation method 3 is described in detail. The generation of virtual pressing force based on the virtual force generation method 3 is performed by the virtual force learning unit 55. The virtual force learning unit 55 has the following functions: extracting useful rules, knowledge expressions, judgment criteria, etc. present in the set of input data through analysis and outputting its judgment result, and performing knowledge learning (machine learning). There are various machine learning methods, but if they are roughly divided, they can be divided into "supervised learning", "unsupervised learning" and "reinforcement learning", for example. In addition, on the basis of implementing these methods, there is also a method called "deep learning" that learns the extracted feature value itself. In the present embodiment, it is assumed that "supervised learning" is applied to the machine learning of the virtual force learning unit 55.

[0080] As described in the above-mentioned "virtual force generation method 2", it is considered that the relative distance between the tool tip position and the workpiece, the relative speed, the coefficient related to the rigidity or dynamic friction of the target workpiece, the coefficient related to the rigidity of the tool, etc., are correlated with the magnitude of the pressing force when the tool tip is in contact with the target workpiece. Therefore, the virtual force learning unit 55 performs learning using learning data that uses these values ​​that are correlated with the magnitude of the pressing force as input data and the pressing force detected by the force sensor in this case as answer data.

[0081] As a specific example of learning model construction, there may be an example of constructing a learning model corresponding to the first to third calculation examples of the virtual force F described above. In the case of constructing a learning model corresponding to the first calculation example of the virtual force F described above, learning data is collected with the relative distance (δ) between the tool tip position and the target workpiece, the relative speed (Vc), and the values ​​(Kd, Kc) related to the rigidity of the target workpiece (or at least the relative distance (δ) between the tool tip position and the target workpiece and the value (Kd) related to the rigidity of the workpiece) as input data, and the pressing force detected by the force sensor in this case as answer data. Then, learning is performed using the learning data to construct a learning model.

[0082] In the case of constructing a learning model corresponding to the second calculation example of the virtual force F, learning data is collected with the movement amount (δ) of the tool front end position and the "virtual spring coefficient of the tool" as input data, and the pressing force detected by the force sensor in this case as answer data. Then, learning is performed using the learning data to construct the learning model. In addition, learning data (training data) consisting of input data and answer data as the pressing force detected by the force sensor in this case can also be collected, and learning is performed using the learning data to construct the learning model, wherein the input data includes at least one of a coefficient related to the rigidity of the target workpiece and a coefficient related to the rigidity of the tool part (tool), and a distance (δ) of the tool part relative to the target workpiece when the tool part is in contact with the target workpiece.

[0083] In the case of constructing a learning model corresponding to the third calculation example of the virtual force F, learning data is collected, which has the moving position (Tx) based on the speed command and the position (d) of the tool tip after the actual movement in response to the speed command as input data, and the pressing force detected by the force sensor in this case as answer data. Then, learning is performed using the learning data to construct a learning model. The learning in this case is equivalent to the action of learning the coefficient k.

[0084] The learning described above can be achieved using a neural network (e.g., a three-layer neural network). The action modes of the neural network include a learning mode and a prediction mode. In the learning mode, the learning data (input data) described above is assigned as an input variable to the neural network, and the weights applied to the inputs of each neuron are learned. The error between the output value when the input data is assigned to the neural network and the correct value (answer data) is taken, and the error is backpropagated to each layer of the neural network, and the weights of each layer are adjusted so that the output value is close to the correct value, thereby performing weight learning. When a learning model is constructed through such learning, the virtual pressing force can be predicted using the input data described above as an input variable.

[0085] The sound output unit 74 outputs a sound that indicates, by volume, the magnitude of the virtual force generated by the virtual force generator 54. For example, during the execution of the force control simulation, a sound corresponding to the magnitude of the virtual force generated by the virtual force generator 54 is output in real time, thereby enabling the operator to more intuitively grasp the magnitude of the virtual force.

[0086] Next, the grinding amount estimation method 2 is described. The learning in the grinding amount estimation method 2 is performed by the grinding amount learning unit 57. As described above, the robot's motion trajectory, motion speed, and pressing force (virtual pressing force) are respectively correlated with the grinding amount. The grinding amount learning unit 57 constructs a learning model corresponding to these parameters and the grinding amount through machine learning. Here, "supervised learning" is applied as machine learning.

[0087] Learning in this case can be constructed using a neural network (e.g., a three-layer neural network), for example. In the learning mode, the above-mentioned learning data (the robot's motion trajectory, motion speed, and virtual pressing force) are given as input variables to the neural network, and the weights applied to the inputs of each neuron are learned. The error between the output value when the neural network is given the input data and the correct value (answer data; grinding amount) is taken, and the error is backpropagated to each layer of the neural network, and the weights of each layer are adjusted so that the output value is close to the correct value, thereby performing weight learning. When a learning model is constructed through such learning, the robot's motion trajectory, motion speed, and virtual pressing force can be used as input to estimate the grinding amount.

[0088] The control device 50 (grinding amount estimation unit 56) causes an image representing a virtual pressing force generated using the above-mentioned virtual force generation methods 1 to 3 and the grinding amount estimated using any of the above-mentioned grinding amount estimation methods 1 and 2 to be displayed on the display device 70 as an augmented reality image or a virtual reality image. The control device 50 (grinding amount estimation unit 56) provides the display device 70 with information representing the size and generation location of the virtual pressing force obtained by performing a force control simulation of the grinding operation, as well as the estimated result of the grinding amount (grinding position and grinding amount). The AR / VR image processing unit 72 of the display device 70 causes the image representing the virtual pressing force and the estimated grinding amount to be superimposed on the image in the real space or the position corresponding to their generation location in the image in the virtual space for display. In the case of generating a virtual reality image, for example, it can also be set as follows: the control device 50 provides the display device 70 with model data and configuration position information of each object in the working space including the manipulator 10. In addition, the display device 70 has a position sensor (optical sensor, laser sensor, magnetic sensor) and an acceleration sensor (gyro sensor) for obtaining the position of the display device 70 in the working space, so that the relative position relationship between the coordinate system (camera coordinate system) fixed to the display device and the world coordinate system fixed to the working space can be grasped.

[0089] Fig. 9 and Fig.10 An example of an image representing a virtual pressing force is shown. Fig. 9The display example of an image showing a virtual pressing force is shown in the case where the teaching trajectory L91 is a trajectory that is relatively close to the surface of the target workpiece W1 near the protrusion 82. Fig. 9 In the case of , the virtual pressing force is represented by the arrow image 191, and it is represented that the virtual pressing force becomes larger in the vicinity of the protrusion 82 on the surface of the target workpiece W1 where the teaching track L91 is relatively close. Fig.10 The display example of the image showing the virtual pressing force is shown in the case where the teaching trajectory L92 is a trajectory in which the vicinity of the protrusion 82 is relatively far from the surface of the target workpiece W1. Fig.10 In the case of , the virtual pressing force is represented by the arrow image 192, showing that the virtual pressing force becomes smaller in the vicinity of the protrusion 82 which is relatively far from the surface of the target workpiece W1 along the teaching track L92.

[0090] Next, refer to Fig.11 and Fig.12 An example of an augmented reality image in which an image showing a virtual pressing force and an estimated grinding amount is superimposed on a real image is described below. Fig.11 In the example of FIG. 1 , an image L93 showing the teaching trajectory, an image 193 showing the position and magnitude of the virtual pressing force generated by the length of the arrow, and an image 211 showing the estimated grinding amount are superimposed and displayed on a real image including the grinding stone 9 and the target workpiece W1. Fig.11 In the example of the image, it can be understood that the virtual pressing force is relatively larger in the area of ​​the protrusion 82, and the estimated grinding amount in the area of ​​the protrusion 82 is larger than the estimated grinding amount in the area without the protrusion 82. In addition, these images L93 representing the teaching track, the image 193 representing the virtual pressing force, and the image 211 representing the estimated grinding amount are made into images representing a three-dimensional area. In this case, the operator can visually grasp the virtual pressing force and the estimated grinding amount from a desired line of sight direction by moving the line of sight.

[0091] Fig.12 Indicates that it will include Fig.11 The image L93 showing the teaching track, the image 193 showing the position and size of the virtual pressing force by the length of the arrow, and the image 121 showing the estimated grinding amount are arranged next to the target workpiece W1 as an augmented reality image and superimposed on the real image. For example, there is also the following case: when the amount of information to be provided as an augmented reality image is large, such as Fig.12 In that way, the augmented reality image can be displayed next to the real object in the image, which is more convenient for the operator (wearer).

[0092] The recommended value generating unit 58 has a function of displaying advice as an image showing how to adjust the estimated grinding amount by adjusting the operation trajectory, operation speed, force control gain, etc. based on the comparison result between the estimated grinding amount and the grinding amount reference value indicating the desired grinding amount. Fig.13 An example is shown below: Fig.11 When the estimated grinding amount (image 211) shown is compared with the grinding amount reference value, since the estimated grinding amount in the region of the protrusion 82 exceeds the grinding amount reference value, the image L101 representing the recommended trajectory is displayed to reduce the estimated grinding amount in the region. The distance from the target workpiece W1 in the vicinity of the protrusion 82 in the image L101 of the recommended trajectory is longer than that in the case of the image L93 representing the teaching trajectory. Thus, in the case of the recommended trajectory (image L101), the estimated grinding amount can be suppressed within the grinding amount reference value.

[0093] Fig.14 An example is shown in which recommended values ​​related to the motion speed of a robot (tool) are presented. Fig.14 In the example of FIG. 1 , an image showing the teaching speed is displayed next to the teaching track (image L93), and images 102 showing the recommended speed are displayed for each section obtained by dividing the teaching track (image L93). Fig.14 In the example of FIG. 1 , the teaching speed for the teaching track (image L93) is 50 mm / s, whereas the recommended speed in the region of the protrusion 82 of the target workpiece W1 is 70 mm / s, and the recommended speed in the region other than the protrusion 82 is 50 mm / s. In the case of this recommended speed, the speed in the region of the protrusion 82 is higher than the teaching speed, so the estimated grinding amount is reduced and converges within the grinding amount reference value.

[0094] The recommended value generating unit 58 may be able to specify parameters to be used for adjustment, for example, by means of an operating unit of the control device 50. For example, when the operator does not want to change the teaching trajectory due to concerns about an increase in cycle time, the recommended value generating unit 58 may be configured such that parameters other than the teaching trajectory (teaching speed, target pressing force, etc.) can be specified as parameters to be adjusted.

[0095] The recommended value generating unit 58 compares the estimated grinding amount with the grinding amount reference value. For example, when the estimated grinding amount is larger than the grinding amount reference value, the action trajectory, action speed, force control parameters, etc. are adjusted in the direction of reducing the estimated grinding amount and confirmed by executing a force control simulation, thereby achieving the above-mentioned recommended value.

[0096] The grinding amount estimation unit 56 may also be configured to calculate the area of ​​the portion of the target workpiece that has been ground using the grinding material (hereinafter referred to as the grinding area). Even when the same grinding material is used, the grinding area varies depending on the angle of the grinding material relative to the target workpiece. Fig.15 The grinding area SA1 when the grinding tool 119 is brought into contact with the target workpiece W51 in an upright posture relative to the target workpiece W51 to grind the target workpiece W51 is the same as that in the case of Fig.16 As shown in the figure, when the grinding tool 119 is brought into contact with the target workpiece W51 in a posture in which the grinding tool 119 is laid down relative to the target workpiece W51 to grind the target workpiece W51, the grinding area SA2 is larger.

[0097] The polishing amount estimation unit 50 calculates the polishing area as follows. Fig.17 As shown in FIG. 1 , it is assumed that the grinding area S when the grinding material 9 is brought into contact with the target workpiece W51 is obtained. L In this case, it is assumed that the pressing force from the tool (abrasive material 9) on the target workpiece, the rotation amount of the tool (abrasive material 9), the material, and the moving speed do not change. In addition, it is assumed that even if the tool (abrasive material 9) is tilted, the overall grinding amount does not change. The volume of the part cut by grinding is V, and the movement amount of the tool (abrasive material 9) is Fig.17 The amount of movement in the depth direction of the drawing is d, and the tool tip angle is a. L represents the length of the inclined surface after cutting. In this case, the following equation holds true.

[0098] V / d=(1 / 2)·Lsin(a)·Lcos(a)

[0099] Assuming that V / d is constant, the above equation is transformed as follows.

[0100] 2V / d=L 2 sin(a)·cos(a)

[0101] 2V / d=(L 2 sin(2a)) / 2

[0102] Based on the above, the length L of the cut surface is calculated as follows.

[0103] L=(4V / (d / sin(2a))) 1 / 2

[0104] The grinding area can be obtained by multiplying the length L by the movement amount d of the tool.

[0105] The tool selection unit 59 has a function of accepting a selection made by a user from a plurality of types of tools stored in advance through the operation unit of the control device 50, or a function of automatically selecting an appropriate tool and posture from a plurality of types of tools stored in advance based on information such as grinding amount, grinding area, cycle time, etc. The force control simulation execution unit 52 virtually mounts the tool selected by the tool selection unit 59 on the manipulator 10 to execute a force control simulation. Fig.18 An example of a grinding tool and a posture selected when a relatively wide area is required or allowed as the grinding area of ​​the target workpiece W61 is shown. Fig.18 In the example of FIG. 5 , the grinding tool has a long shape and is used in a posture laid down relative to the target workpiece W61, thereby ensuring a wide grinding area.

[0106] Fig.19 An example of a grinding tool and a posture selected when a relatively narrow area is required as the grinding area of ​​the target workpiece W62 is shown. Fig.19 In the example shown, a relatively short grinding tool is selected, and the grinding tool is in an upright position relative to the target workpiece W62. In this case, the grinding area can be narrowed.

[0107] Fig. 20 An example of a grinding tool and a posture selected when reducing the grinding amount and grinding area is shown. Fig. 20 In the example, a grinding tool 221 having a metal brush is selected as the grinding material, and is positioned so that the peripheral surface of the grinding material is in contact with a portion of the burr 181 on the target workpiece W63 (the center line of the grinding tool is tilted by about 45 degrees relative to the vertical direction).

[0108] In addition, for the relationship between the material of the grinding tool, the material of the target workpiece and the grinding amount, the correlation can be obtained based on the following considerations. On the grinding tool side, the roughness of the abrasive has a stronger relationship with the grinding amount than the rigidity, so the grinding amount can also be predicted by obtaining measured data according to the roughness of the abrasive. For example, the approximate model of the grinding amount is set in a manner that the greater the roughness of the abrasive, the greater the grinding amount. Regarding the workpiece, there are workpieces that are difficult to cut depending on the material of the workpiece. Therefore, as an index of the rigidity of the material, there is a Young's modulus representing ductility and a plasticity coefficient representing plasticity, but these coefficients can also be used directly as coefficients representing the grinding amount, or these coefficients can be set to (1 / Young's modulus) and the like to be used as coefficients representing the grinding amount. Alternatively, in the case of the workpiece, measured data is obtained for the cutting amount corresponding to the rigidity of the material and an approximate model is obtained to predict the grinding amount.

[0109] As described above, according to the present embodiment, the operator can intuitively grasp the estimated grinding amount and can easily adjust the teaching trajectory and the force control parameter.

[0110] The present invention has been described above using typical embodiments, but it will be apparent to those skilled in the art that changes to the above-described embodiments and various other changes, omissions, and additions may be made without departing from the scope of the present invention.

[0111] The distribution of functions among the control device 50, the display device 70, and the external computer 90 in the above-mentioned embodiment is an example, and the arrangement of these functional blocks can be changed. The imaging device may be arranged at a fixed position in the work space as a separate device from the display device.

[0112] The functional blocks of the control device and the display device may be realized by the CPU of these devices executing various software stored in the storage device, or may be realized by a configuration mainly based on hardware such as an ASIC (Application Specific Integrated IC).

[0113] The program for executing the various simulation processes in the above-described embodiments can be recorded in various computer-readable recording media (eg, semiconductor memories such as ROM, EEPROM, and flash memory, magnetic recording media, and optical disks such as CD-ROM and DVD-ROM).

[0114] Description of Reference Numerals

[0115] 3: force sensor; 10: robot manipulator; 11: tool unit; 50: control device; 51: storage unit; 52: force control simulation execution unit; 53: robot motion control unit; 54: virtual force generator; 55: virtual force learning unit; 56: grinding amount estimation unit; 57: grinding amount learning unit; 58: recommended value generation unit; 59: tool selection unit; 70: display device; 71: camera device; 72: AR / VR image processing unit; 73: display; 74: sound output unit; 90: external computer; 91: physical simulation unit; 100: robot system.

Claims

1. A grinding amount estimation device for estimating a grinding amount in a grinding operation performed by bringing a grinding tool mounted on a robot manipulator into contact with a target workpiece by force control, the grinding amount estimation device comprising: a storage unit storing an operation program; and a grinding amount estimating unit that estimates the grinding amount based on at least one of a motion trajectory of the grinding tool, a motion speed of the grinding tool, and a pressing force of the grinding tool on the target workpiece obtained according to the motion program, in, The storage unit further stores a force control parameter as a parameter related to the force control. The grinding amount estimation device further includes a force control simulation execution unit that executes a simulation of the force control based on the operation program and the force control parameter. The force control simulation execution unit includes a virtual force generation unit that virtually generates a pressing force applied from the grinding tool to the target workpiece when the grinding tool is in contact with the target workpiece, based on position information of the grinding tool obtained from a simulation result of the force control.

2. The grinding amount estimation device according to claim 1, wherein: The force control simulation execution unit obtains the motion trajectory, the motion speed, and the pressing force based on the position information of the grinding tool obtained from the simulation result of the force control.

3. The grinding amount estimation device according to claim 1, wherein: The device further comprises a physical simulation unit that uses a motion equation representing the robot manipulator and performs a physical simulation of the motion of the robot manipulator based on the force control parameter. The virtual force generation unit obtains the pressing force based on the position information of the grinding tool obtained by the physical simulation in a state where the grinding tool is in contact with the target workpiece.

4. The grinding amount estimation device according to claim 3, wherein: The virtual force generating unit obtains the pressing force based on any one of a coefficient related to the rigidity of the target workpiece, a coefficient related to the rigidity of the grinding tool and a spring coefficient of the grinding tool, and a distance of the grinding tool relative to the target workpiece when the grinding tool is in contact with the target workpiece.

5. The grinding amount estimation device according to any one of claims 1 to 4, wherein: The grinding amount estimation unit estimates the grinding amount based on any one of a model obtained by linearly approximating or curvilinearly approximating the correlation between the movement trajectory of the grinding tool and the grinding amount based on measured data, a calculation model obtained by linearly approximating or curvilinearly approximating the correlation between the movement speed of the grinding tool and the grinding amount based on measured data, and a calculation model obtained by linearly approximating or curvilinearly approximating the correlation between the pressing force of the grinding tool and the grinding amount based on measured data.

6. The grinding amount estimation device according to any one of claims 1 to 4, wherein: The invention further comprises a grinding amount learning unit for performing machine learning based on training data composed of input data including a motion trajectory of the grinding tool, a motion speed of the grinding tool, and a pressing force of the grinding tool on the target workpiece, and answer data corresponding to the input data as an actually measured grinding amount. The polishing amount estimation unit estimates the polishing amount using a learning model constructed by machine learning of the polishing amount learning unit.

7. The grinding amount estimation device according to any one of claims 1 to 4, wherein: Also available: an imaging device that acquires an image of a real working space including the robot manipulator and the target workpiece; and A display device superimposes an image representing the estimated grinding amount on an image of the work space as an augmented reality image.

8. The grinding amount estimation device according to any one of claims 1 to 4, wherein: The storage unit further stores model data representing shapes of the robot manipulator, the grinding tool, and the target workpiece, and information on their arrangement positions. The grinding amount estimation device further includes a display device that uses the model data and the information on the configuration position to display an image showing the estimated grinding amount by superimposing it on a virtual reality image configured in a virtual work space including the grinding tool and the target workpiece.

9. The grinding amount estimation device according to claim 7, wherein: further comprising a recommended value generating unit for generating a recommended adjustment value related to the teaching trajectory or the force control parameter based on a comparison result between the estimated grinding amount and a predetermined grinding amount reference value, The display device further displays an image indicating the recommended adjustment value by superimposing it on the image of the actual work space.

10. The grinding amount estimation device according to any one of claims 2 to 4, wherein: The invention further comprises a tool selection unit for selecting a grinding tool to be used from a plurality of types of grinding tools based on information indicating a required grinding amount or a required grinding area. The force control simulation execution unit virtually mounts the selected grinding tool on the robot manipulator to execute the force control simulation.

11. The grinding amount estimation device according to claim 8, wherein: further comprising a recommended value generating unit for generating a recommended adjustment value related to the teaching trajectory or the force control parameter based on a comparison result between the estimated grinding amount and a predetermined grinding amount reference value, The display device further displays an image representing the recommended adjustment value by superimposing it on the virtual reality image.

Images