JP7611763B2 - Simulation device and simulation method - Google Patents

Description

The present invention relates to a simulation device and a simulation method.

In recent years, the use of robots has been promoted as an alternative to labor shortages caused by the declining birthrate and aging population, and to work in dangerous areas due to natural disasters. In order to expand the range of robots' operational areas, robots are required to be able to manipulate flexible objects such as cables, hoses, and fabrics. For example, robots that take over recovery work in dangerous areas due to natural disasters are expected to have the task of connecting power plugs and fire hydrant hoses to plugs. Additionally, nursing care robots are required to manipulate clothing to assist with dressing.

When a flexible object is manipulated by a robot, the shape of the flexible object changes in response to the manipulation. Therefore, when planning a method for manipulating a flexible object by a robot, the plan needs to be based on an estimate of the shape change of the flexible object. In this case, if the accuracy of the estimation is low, the robot operating based on the planned content of the manipulation method may not be able to perform the desired manipulation of the flexible object, and the task may fail. Furthermore, if the accuracy of the estimation of the shape change of the flexible object is low, there is a possibility that the flexible object may interfere with the surrounding environment or the robot itself, compromising safety.

For example, Non-Patent Document 1 discloses a technique for performing high-speed analysis by approximating the deformation of a flexible object on a two-dimensional plane. However, when the moving speed of the flexible object is not constant or when the flexible object undergoes complex deformation such as twisting, it is expected that the technique described in Non-Patent Document 1 will result in a decrease in analysis accuracy.

Patent document 1 discloses a technology that estimates the three-dimensional behavior of a flexible object by automatically adjusting the spring constant, damping coefficient, and restitution coefficient of an analytical model so that the behavior of the actual flexible object matches the results of a physical simulation.

P. Chang and T. Padir. “Sim2real2sim: Bridging the gap between simulation and real-world in flexible object manipulation.” 2020 Fourth IEEE International Conference on Robotic Computing, arXiv:2002.02538, 2020. JP 2013-035083 A

However, when estimating the behavior of a flexible object through physical simulation, errors occur between the behavior of the actual object and that of the analytical model due to discretization errors when reproducing the actual flexible object in an analytical model, and approximations made to speed up analytical algorithms. Even when obtaining the behavior of an actual flexible object through measurement, errors occur between the behavior of the actual object and that of the analytical model due to sensor-specific errors and external disturbances.

If a robot is operated based on estimates that are based on such errors, there is a possibility that dangerous actions will be performed, even though safety should be guaranteed. In order to prevent such situations from occurring, it is considered effective to evaluate the probabilistic distribution of the behavior of an analytical model of a flexible object. In other words, the distribution that shows the correspondence between the values taken by the coefficients of an analytical model of a flexible object and the probability of taking those values needs to be evaluated probabilistically.

However, the technology described in Patent Document 1 does not perform such evaluation, so there is a risk that the accuracy of estimating the behavior of flexible objects may be low.

The present invention was made in consideration of the above situation, and its purpose is to make it possible to probabilistically evaluate the behavior of flexible objects.

A simulation device according to one embodiment of the present invention comprises a memory unit that stores measurement results of the dynamic displacement of a flexible object in the time direction as the behavior of the flexible object when a predetermined operation is performed on an operation unit connected to the other end of the flexible object , the flexible object being an amorphous object whose one end is fixed to a boundary point and whose other end is fixed to a boundary point; a simulation unit that analyzes the behavior of an analytical model of the flexible object when the operation is performed by physical simulation; a parameter estimation unit that modifies model parameters of the analytical model so that the behavior of the flexible object stored in the memory unit matches the behavior of the analytical model of the flexible object analyzed by the simulation unit, and estimates the probability distribution of the modified model parameters; and an interference evaluation unit that calculates the probability of interference between the analytical model of the flexible object and an analytical model to be evaluated whose interference with the flexible object is evaluated by analyzing the behavior of the flexible object by the simulation unit . The parameter estimation unit estimates the probability distribution of the model parameters using a data assimilation technique, the simulation unit sets external forces according to the time history of the dynamic displacement of the flexible object read out from the memory unit in the analytical model of the operation unit, and analyzes the behavior of the flexible object based on the probability distribution of the model parameters, and the interference evaluation unit calculates the interference probability by performing a physical simulation when a specified operation is performed on the flexible object using multiple model parameters obtained by generating random numbers based on the probability distribution of the model parameters, and evaluates that the analytical model of the flexible object and the analytical model of the evaluation object interfere with each other when the area occupied by the analytical model of the evaluation object in the virtual space where the physical simulation is performed, or the interference detection area set in the analytical model of the evaluation object overlaps with the area occupied by the analytical model of the flexible object.

Furthermore, a simulation method according to one aspect of the present invention is a simulation method using a simulation device having a memory unit that stores measurement results of dynamic displacement in the time direction of a flexible object as behavior of the flexible object when a predetermined operation is performed on an operation unit connected to an operation unit connected to one end of the flexible object, the flexible object being an indefinite object whose shape is not easily determined and whose one end is fixed to a boundary point, the method including a parameter estimation unit that constructs an analytical model of the flexible object, a parameter estimation unit that analyzes the behavior of the analytical model of the flexible object when the operation is performed by using a physical simulation, a parameter estimation unit that modifies model parameters of the analytical model so that the behavior of the flexible object stored in the memory unit matches the behavior of the analyzed analytical model of the flexible object and estimates a probability distribution of the modified model parameters, and an interference evaluation unit that evaluates interference between the analytical model of the flexible object and the flexible object by the simulation unit analyzing the behavior of the flexible object. and a procedure for calculating the probability of interference between the analytical model of the flexible object and the analytical model of the object , wherein the parameter estimation unit estimates the probability distribution of the model parameters using a data assimilation technique, the simulation unit sets an external force that follows the time history of the dynamic displacement of the flexible object read out from the memory unit in the analytical model of the operation unit and analyzes the behavior of the flexible object based on the probability distribution of the model parameters, and the interference evaluation unit calculates the interference probability by performing a physical simulation when a predetermined operation is performed on the flexible object using a plurality of model parameters obtained by generating random numbers based on the probability distribution of the model parameters, and evaluates that the analytical model of the flexible object and the analytical model of the object interfere with each other when an area occupied by the analytical model of the flexible object in the virtual space in which the physical simulation is performed or an interference judgment area set in the analytical model of the object overlaps with an area occupied by the analytical model of the flexible object .

According to at least one aspect of the present invention, it becomes possible to probabilistically evaluate the behavior of flexible objects.
Problems, configurations and effects other than those described above will become apparent from the following description of the embodiments.

1 is a diagram illustrating a configuration of a simulation device according to a first embodiment of the present invention. 1 is a block diagram showing an example of a hardware configuration of each device included in a simulation device according to a first embodiment of the present invention; FIG. 2 is a diagram illustrating an example of the configuration of a virtual space according to the first embodiment of the present invention. 4 is a graph showing an example of estimating model parameters using an ensemble Kalman filter according to the first embodiment of the present invention. 5 is a graph showing an example of acquisition of model parameters by an interference assessment unit according to the first embodiment of the present invention. 5 is a diagram showing an example of a method for evaluating interference between a flexible object and an evaluation target by an interference evaluation unit according to the first embodiment of the present invention; FIG. 3 is a flowchart showing an example of a procedure of a simulation method by the simulation device according to the first embodiment of the present invention. 13 is a diagram showing an example of evaluation of interference between a flexible object and an evaluation target by an interference evaluation unit according to a first modified example of the first embodiment of the present invention; FIG. 13 is a diagram showing an example of evaluation of interference between a flexible object and an evaluation target by an interference evaluation unit according to a second modified example of the first embodiment of the present invention; FIG. FIG. 13 is a diagram illustrating an example of the configuration of a simulation device according to a second embodiment of the present invention. FIG. 11 is a diagram showing an example of an operation plan for a robot according to Example 1 of the second embodiment of the present invention. FIG. 13 is a diagram showing an example of a result of a physical simulation in a case where the operation unit is moved at a constant speed, in Example 1 of the second embodiment of the present invention. FIG. 13 is a diagram showing an example of a result of a physical simulation in a case where the moving speed of the operation part is decelerated at a position (a route position) near an evaluation target, according to Example 1 of the second embodiment of the present invention. FIG. 11 is a diagram showing an example of the planning of an operation plan for a robot at node n by a planner, according to a second example of the second embodiment of the present invention.

Below, examples of modes for carrying out the present invention (hereinafter referred to as "embodiments") will be described with reference to the attached drawings. The present invention is not limited to the embodiments, and the various numerical values in the embodiments are merely examples. Furthermore, in this specification and drawings, identical components or components having substantially the same functions will be given the same reference numerals, and duplicate explanations will be omitted.

First Embodiment
[Configuration of the simulation device]
FIG. 1 is a diagram for explaining the configuration of a simulation device 10 according to a first embodiment of the present invention. The simulation device 10 is a device for simulating the behavior of a flexible object when the flexible object is operated by a robot (not shown). A flexible object is an object that is soft and has an indefinite shape that is difficult to determine in a fixed manner, and examples of the flexible object include cables, wires, hoses, cloth, and sheets. Note that the flexible object may be an object whose material is not soft, but whose movement caused by an operation applied to one end of the object is transmitted to the other end, for example, an object in which multiple rod-shaped objects are connected via joints.

As shown in FIG. 1, the simulation device 10 includes a memory unit 11, an analytical model construction unit 12, a simulation unit 13, a parameter estimation unit 14, an interference evaluation unit 15, a display control unit 16, and a display unit 17.

The memory unit 11 stores the behavior of a flexible object when a specified operation is performed on the flexible object, i.e., the measurement results of the dynamic displacement of the flexible object in the time direction. When a specified operation is performed on a flexible object, the behavior of the flexible object may include the flexible object's oscillation, twisting, or stillness, and therefore the dynamic displacement of the flexible object stored in the memory unit 11 may also include these states.

In this embodiment, it is assumed that there is an evaluation object that the flexible object may collide with, get caught on, or wrap around, so that a contact operation with the evaluation object that is expected to interfere with the flexible object is also performed as a specified operation on the flexible object.

The dynamic displacement of a flexible object can be measured, for example, by performing an operation on a flexible object on which markers (feature points) are placed at multiple locations, analyzing video images of the manipulated flexible object captured by a high-speed camera or the like, and acquiring time-dependent changes in the coordinates and moving speed of each marker. Alternatively, the displacement of the flexible object can be measured directly using a sensor such as a strain sensor.

Measurement methods using cameras are more susceptible to disturbances than methods that directly measure displacement, but have the advantage that measurements can be made in a short time because the measuring equipment is easy to install. The optimal measurement method varies depending on the application conditions of the present invention, but any measurement method can be applied. For example, if the robot is an autonomous robot and needs to perform measurements by itself, it may be difficult to attach a sensor to a flexible object. In such cases, it is desirable to perform simple measurements using a camera.

In this embodiment, in order to statistically treat the measurement data, the dynamic displacement of the flexible object is measured multiple times, and the average and variance of the dynamic displacement are calculated. In other words, the memory unit 11 stores the average and variance of the measurement data of the flexible object as the dynamic displacement of the flexible object. Alternatively, instead of performing the measurement multiple times, it is also possible to empirically determine the measurement error based on prior information about the sensor.

Measurement of the dynamic displacement of the flexible object and calculation of the average and variance of the measurement data may be performed when the simulation is performed by the simulation device 10, or may be performed in advance before the simulation.

The analytical model construction unit 12 constructs an analytical model used when the simulation unit 13 executes a simulation (physical simulation). Specifically, the analytical model construction unit 12 constructs a virtual space used when the simulation unit 13 simulates the behavior of a flexible object. The analytical model construction unit 12 makes an analytical model of the flexible object to be operated and an analytical model related to an object to be evaluated for interference with the flexible object (both are three-dimensional models) appear in the virtual space in which the simulation is performed. The analytical model construction unit 12 then sets a computational grid to be used in the physical simulation for the three-dimensional model, and divides it into multiple elements. An example of the configuration of the virtual space by the analytical model construction unit 12 will be described in detail with reference to FIG. 3 described later.

The simulation unit 13 performs a physical simulation using the analytical model constructed by the analytical model construction unit 12 and analytical parameters acquired from the parameter estimation unit 14 described below. One method of physical simulation performed by the simulation unit 13 is a method of analyzing a continuous elastic body model based on physical laws, such as the finite element method. Other methods include a method of approximating a model or algorithm, such as a spring-mass model or PBD (Position Based Dynamics).

Simulations using the spring-mass model, PBD, etc. are physically inaccurate compared to the finite element method, but the time required for analysis is shorter compared to the finite element method. The optimal analysis method varies depending on the application conditions of the present invention, but any analysis method can be applied in this embodiment. For example, in a public place or disaster site, in a situation where an autonomous robot only operates a target flexible object a few times, the speed of analysis is more important than the accuracy of the simulation. Therefore, in such cases, it is preferable to use an approximation model such as the spring-mass model or PBD as a physical simulation method.

The parameter estimation unit 14 modifies the model parameters of the analytical model so that the dynamic displacement of the flexible object read from the memory unit 11 coincides with the dynamic displacement of the flexible object obtained by a physical simulation using the analytical model constructed by the analytical model construction unit 12. Then, the parameter estimation unit 14 estimates the probability distribution of the modified model parameters.

The model parameters include at least the elastic coefficient of the flexible object. Depending on the application conditions of the present invention, the model parameters can also include other material parameters of the flexible object (such as damping coefficient and density), boundary conditions, and coefficients related to the contact force (friction force, normal force) between the flexible object and the object to be evaluated.

In this embodiment, the parameter estimation unit 14 estimates these multiple model parameters simultaneously by using a data assimilation method. Data assimilation methods include a Kalman filter or particle filter that assimilates sequentially, and an adjoint method that assimilates non-sequentially. In this embodiment, a case where an ensemble Kalman filter is applied will be described, but any data assimilation method can be applied to the present invention. The estimation of model parameters using an ensemble Kalman filter will be described in detail with reference to FIG. 4 below.

The interference assessment unit 15 calculates the probability of interference between the flexible object and the evaluation target by analyzing the behavior of the flexible object when an arbitrary operation is performed by the simulation unit 13, based on the probability distribution of the model parameters estimated by the parameter estimation unit 14. The method of calculating the probability of interference between the flexible object and the evaluation target by the interference assessment unit 15 will be described in detail with reference to Figures 5 and 6 below.

The display control unit 16 controls the display unit 17 to display the results of the physical simulation performed by the simulation unit 13, the interference probability calculated by the interference evaluation unit 15, etc.

The display unit 17 is a monitor configured with an LCD (Liquid Crystal Display) or organic electroluminescence, and displays characters, drawings, moving images, etc. on the screen based on the control of the display control unit 16.

[Computer hardware configuration]
Next, the hardware configuration of each device for realizing the functions of the simulation device 10 shown in Fig. 1 will be described with reference to Fig. 2. Fig. 2 is a block diagram showing an example of the hardware configuration of each device included in the simulation device 10.

Calculator 50 is hardware used as a so-called computer. Calculator 50 includes a CPU (Central Processing Unit) 501, a ROM 502, and a RAM (Random Access Memory) 503, each of which is connected to bus B. Calculator 50 further includes a display unit 504, an operation input unit 505, a non-volatile storage 506, and a network interface 507.

The CPU 501 reads out the program code of the software that realizes each function according to this embodiment from the ROM 502, expands it in the RAM 503, and executes it. Note that the calculator 50 may include a processing device such as an MPU (Micro-Processing Unit) instead of the CPU 501. Variables, parameters, etc. that are generated during the calculation processing by the CPU 501 are temporarily written to the RAM 503.

The display unit 504 is, for example, a monitor configured with an LCD or the like, and displays the results of the processing performed by the computer 50. The function of the display unit 17 of the simulation device 1 according to this embodiment is realized by the display unit 504.

The operation input unit 505 is a block into which operations by the user are input, and is configured, for example, with a keyboard, a mouse, a touch sensor, etc. The display unit 504 and the operation input unit 505 may be configured integrally as a touch panel.

For example, the non-volatile storage 506 may be a hard disk drive (HDD), a solid state drive (SSD), a flexible disk, an optical disk, a magneto-optical disk, a CD-ROM, a CD-R, a magnetic tape, a non-volatile memory card, etc. In addition to the operating system (OS) and various parameters, the non-volatile storage 506 also records programs for operating the computer 50. The ROM 502 and the non-volatile storage 506 record programs, data, etc. required for the CPU 501 to operate, and are used as an example of a computer-readable non-transient storage medium that stores a program executed by the computer 50.

The network interface 507 may be, for example, a network interface card (NIC), and can transmit and receive various types of data to and from external devices via a network such as a LAN or a communication line.

[Virtual space configuration]
Next, the virtual space constructed by the analytical model construction unit 12 will be described. Fig. 3 is a diagram showing an example of the configuration of the virtual space Vs. As shown in Fig. 3, an analytical model Ma11 of a flexible object to be operated and an analytical model Ma21 of an object to be evaluated for interference with the flexible object are provided in the virtual space Vs in which the simulation is performed. In addition, the virtual space Vs is provided with an analytical model Ma31 of an operation unit, which is a location where one end of a flexible object is grasped by a robot (not shown) and where operation by the robot is performed, and an analytical model Ma32 of a boundary point where the other end of the flexible object is fixed.

The simulation unit 13 sets, in each time step of the physical simulation using each analytical model, an external force (forced displacement) that follows the time history of the dynamic displacement of the flexible object read out from the memory unit 11, in the analytical model Ma31 of the operation unit. At this time, the simulation unit 13 sets, in the analytical model Ma31 of the operation unit, an external force that enables the behavior of the analytical model Ma11 of the flexible object to be similar to the dynamic displacement of the flexible object read out from the memory unit 11.

In this embodiment, an example is given in which the boundary point is fixedly supported by a wall or the like (not shown), but the present invention is not limited to this. It is also possible to set the boundary point to be elastically supported, or to set an external force on the analytical model Ma32 of the boundary point. When the boundary point is elastically supported, the parameter estimation unit 14 can estimate the behavior of the boundary point by adding the elastic coefficient of the analytical model Ma32 of the boundary point to the model parameters.

[Model parameter estimation using ensemble Kalman filter]
Next, a method of estimating model parameters using the ensemble Kalman filter by the parameter estimator 14 will be described with reference to Fig. 4. Fig. 4 is a graph showing an example of estimating model parameters using the ensemble Kalman filter.

The vertical axis of the graph shown in FIG. 4 indicates the probability density, and the horizontal axis indicates the state variable. The parameter estimation unit 14 uses an ensemble Kalman filter to estimate the model parameters of the analysis model by the following steps a to d.

Step a: Generate multiple realizations of a state vector consisting of multiple state variables (model parameters and displacements of flexible objects), and define each state vector as an ensemble member Ma. Ensemble members Ma are indicated in Figure 4 by white circles placed on the horizontal axis of the graph.

Step b: The simulation unit 13 performs a physical simulation for each ensemble member Ma and updates the time. That is, it advances (calculates) the time of the physical simulation by Δt (t indicates the unit of time).

Step c: The parameter estimation unit 14 calculates the mean and variance of the ensemble members 41, and expresses the probability density function Dp1 of the state vector (forecast value) before data assimilation based on each calculated value. The probability density function Dp1 is shown by a solid curve in Figure 4.

Step d: The parameter estimation unit 14 sets the probability density function Dp1 of the forecast value as the prior distribution, and sets the probability density function Dp2 obtained from the average and variance of the measured values of the dynamic displacement of the flexible object read from the memory unit 11 as the likelihood. Then, the parameter estimation unit 14 updates the observation by calculating the probability density distribution Dp11 of the state vector (analysis value) after data assimilation based on Bayes' theorem shown in the following equation (1).

p(x|y)∝p(y|x)・p(x)...Equation (1)

In the above formula (1), "p(x|y)" represents the probability density distribution Dp11 of the analytical value, "p(y|x)" represents the probability density function Dp2 of the measured value, and "p(x)" represents the probability density function Dp2 of the forecast value.

By repeating steps b to d described above, the parameter estimation unit 14 causes each value of the ensemble member Ma to converge to a true value. Then, the parameter estimation unit 14 uses the finally obtained ensemble member Ma (which has converged to the true value) to represent the probability distribution of the model parameters.

At this time, the parameter estimation unit 14 can estimate non-homogeneous material parameters of the flexible object by assimilating the material parameters of each element that has been discretized (divided into a finite number of elements) in the flexible object analysis model Ma11 (see FIG. 3) as different state variables.

In this way, when using the method of sequential data assimilation, the procedure of storing information on the dynamic displacement of the flexible object in the memory unit 11 can be executed sequentially or non-sequentially at a stage prior to step d in the procedure of estimating the model parameters using the above-mentioned ensemble Kalman filter.

As described above, when constructing an analytical model, the analytical model construction unit 12 sets a computational grid to be used in a physical simulation for the three-dimensional model and divides the three-dimensional model into multiple elements. In general, when the shape of the three-dimensional model is simplified or when the number of computational grids is insufficient, the accuracy of the analysis decreases.

In contrast, in this embodiment, the parameter estimation unit 14 modifies the model parameters of the analytical model Ma11 of the flexible object so that the actual dynamic displacement of the flexible object read from the memory unit 11 matches the dynamic displacement of the flexible object obtained by physical simulation, and estimates the probability distribution of the modified model parameters. In other words, the parameter estimation unit 14 estimates the model parameters taking into account the roughness of the analytical model, so that a certain degree of roughness of the analytical model is acceptable even when the shape of the three-dimensional model is simplified or the number of calculation grids is insufficient.

Therefore, even if a three-dimensional model cannot be prepared in advance, it is possible to perform a physical simulation using an analytical model Ma11 of a flexible object and an analytical model Ma21 of an evaluation target, which are estimated based on measurement data from a camera, etc. In other words, even when such a physical simulation is performed, it is possible to maintain the accuracy of the estimation of the behavior of the flexible object by the interference evaluation unit 15.

[Method of calculating the probability of interference between a flexible object and an evaluation target]
Next, a description will be given of a method for calculating the probability of interference between a flexible object and an evaluation target by the interference evaluation unit 15. The calculation of the probability of interference between a flexible object and an evaluation target by the interference evaluation unit 15 is performed based on the following procedures A to C.

Step A: The interference assessment unit 15 generates random numbers according to the distribution of the probability density function Dp1 of the model parameters estimated by the parameter estimation unit 14, and obtains the random numbers as a plurality of model parameters Pmi.
Step B: The interference assessment unit 15 performs a physical simulation multiple times using the different acquired model parameters Pmi when an arbitrary operation is performed on the operation unit.
Step C: Based on the results of multiple analyses of the flexible object, the probability that the flexible object will interfere with the evaluation target is calculated.

Figure 5 is a graph showing an example of model parameter acquisition by the interference evaluation unit 15. Figure 5A is a graph showing an example of a method for acquiring model parameters, and Figure 5B is a diagram showing an example of the analysis results of the behavior of a flexible object performed using the model parameters.

The graph on the left side of FIG. 5A shows the correspondence between the probability density in the probability density function Dp1 of the forecast value and the elastic modulus, and the graph on the right side of FIG. 5A shows the correspondence between the probability density in the probability density function Dp1 of the forecast value and the density (kg/m ³ ) of the model parameter.

In the graphs on the left and right of FIG. 5A, the solid curve indicates the probability density function Dp1 of the forecast value, and each of the marks of various shapes arranged on the horizontal axis indicates the model parameter Pmi. In FIG. 5B, the cube on the left indicates the analytical model Ma31 of the operation unit, and the cube on the right indicates the analytical model Ma32 of the boundary point. In addition, each of the multiple solid curves indicates the analysis results Ra of multiple flexible objects, and each of the marks of various shapes arranged on the line of the multiple analysis results Ra indicates the model parameter Pmi.

Figures 5A and 5B show that the higher the probability density, the more model parameters Pmi are obtained, and that the probability density is higher in positions where the analysis results Ra of the flexible object are densely present (positions closer to the average value of the probability density function Dp1).

Figure 6 is a diagram showing an example of a method for evaluating interference between a flexible object and an evaluation target by the interference evaluation unit 15. In Figure 6, the cube on the left side shows an analytical model Ma31 of the operation unit, the cube on the right side shows an analytical model Ma32 of the boundary points, and each of the multiple solid curved lines shows the results Ra of multiple analyses of the flexible object. Also, the circles show the analytical model Ma21 of the evaluation target.

In the evaluation method shown in FIG. 6, when the area occupied by the analytical model Ma21 to be evaluated and the area occupied by the analytical model Ma11 of the flexible object overlap in the virtual space, it is evaluated that there is interference. Note that instead of the area occupied by the analytical model Ma11 to be evaluated, a judgment area may be set around the analytical model Ma11 to be evaluated, and when the judgment area and the area occupied by the analytical model Ma11 of the flexible object overlap, it may be evaluated that there is interference. The evaluation of interference using the judgment area will be described in detail with reference to FIG. 8 below.

In the example shown in Figure 6, among the results of multiple physical simulations shown as the shape of analytical model Ma11 of the evaluation target, the area where analytical model Ma11 of the flexible object and analytical model Ma21 of the evaluation target overlap, i.e., the area where interference occurs, is shown by a dashed curve. Also, in Figure 6, the probability of interference between the flexible object and the evaluation target calculated using the above formula (2) is shown numerically, such as "29%."

The interference evaluation unit 15 defines the number of trials of the physical simulation as N (times), the number of times that interference is determined to occur as M (times), and calculates the probability that the flexible object will interfere with the evaluation target by using the following formula (2).

M/N×100[%]…Formula (2)

In the physical simulation using the above formula (2), the contact force between the flexible object analysis model Ma11 and the evaluation target analysis model Ma21 is not taken into account, so the analysis time is shorter than when the contact force is taken into account. Therefore, when it is desired to shorten the time until the interference evaluation, or when it is necessary to set a large judgment region because the behavior of the evaluation target cannot be predicted, it is preferable to adopt the evaluation using the above formula (2).

The analysis results by the simulation unit 13 and the interference evaluation results by the interference evaluation unit 15 are displayed, for example, on the display unit 17 of the simulation device 10 or on a display unit of a terminal device (not shown) operated by the user. This allows the user to check on the screen the behavior of the flexible object analysis model Ma11 in the physical simulation, the probability of interference between the flexible object analysis model Ma11 and the analysis model Ma21 to be evaluated, etc.

In addition, in the above-described embodiment, an example was given of a case in which there was only one evaluation target for interference with a flexible object, but the present invention is not limited to this. In addition, there may be multiple evaluation targets, and even in the case in which there are multiple evaluation targets, it is possible to perform interference evaluation for each evaluation target.

[Simulation method using a simulation device]
Next, a simulation method performed by the simulation device 10 according to the present embodiment will be described with reference to Fig. 7. Fig. 7 is a flowchart showing an example of the procedure of the simulation method performed by the simulation device 10.

First, the user stores in the memory unit 11 (see FIG. 1) the behavior (dynamic displacement) of the flexible object when a specific operation is performed on the flexible object (step S1). Next, the analysis model construction unit 12 of the simulation device 10 constructs an analysis model that is used when the simulation unit 13 performs a simulation (step S2).

Next, the parameter estimation unit 14 modifies the probability distribution of the model parameters so that the behavior of the flexible object stored in the memory unit 11 matches the behavior of the flexible object in the analysis model Ma11 of the flexible object (step S3). Specifically, as described above, the parameter estimation unit 14 sets the probability density function Dp1 (see FIG. 5) of the forecast value as the prior distribution, and sets the probability density function Dp2 obtained from the average and variance of the measured values of the dynamic displacement of the flexible object in the real environment as the likelihood. Then, the parameter estimation unit 14 modifies the probability distribution of the model parameters by calculating the probability density distribution Dp11 of the state vector after data assimilation based on Bayes' theorem.

Next, the interference evaluation unit 15 analyzes the behavior of the flexible object when an arbitrary operation is performed, and calculates the probability that the flexible object will interfere with the evaluation target (step S4). Specifically, the interference evaluation unit 15 acquires random numbers generated according to the distribution of the probability density function Dp1 of the model parameters estimated (corrected) by the parameter estimation unit 14 as multiple model parameters Pmi, and performs physical simulations multiple times based on the multiple model parameters Pmi. Then, based on the results of the multiple flexible object analyses, it calculates the probability that the flexible object will interfere with the evaluation target. After the processing of step S4, the simulation method by the simulation device 10 ends.

In the above-described embodiment, the parameter estimation unit 14 of the simulation device 10 modifies the model parameters of the analysis model Ma11 of the flexible object so that the behavior of the flexible object stored in the memory unit 11 coincides with the behavior of the analysis model Ma11 of the flexible object analyzed by the simulation unit 13, and estimates the probability distribution of the modified model parameters. Therefore, according to this embodiment, even if errors occur due to discretization performed when performing a physical simulation or approximation aimed at speeding up the analysis algorithm, errors inherent to the sensor that measures the actual behavior of the flexible object, or errors due to disturbances, it is possible to quantitatively (probabilistically) evaluate the errors.

In addition, in the above-described embodiment, the probabilistic distribution of the behavior of the flexible object analysis model Ma11 is estimated, so that even if there are unknown parameters or boundary conditions that are not automatically adjusted to match the actual behavior of the flexible object, it is possible to prevent a decrease in the estimation accuracy of the behavior of the flexible object due to these factors.

In addition, in the above-described embodiment, the parameter estimation unit 14 estimates the probability distribution of the model parameters of the flexible object analysis model Ma11 using a data assimilation technique. Therefore, it is possible to improve the reproducibility of the flexible object analysis model Ma11, thereby improving the estimation accuracy of the behavior of the flexible object.

In the above-described embodiment, the interference evaluation unit 15 calculates the probability of interference between the analysis model Ma11 of the flexible object and the analysis model Ma21 of the evaluation target by analyzing the behavior of the flexible object when the simulation unit 13 performs a predetermined operation based on the probability distribution of the model parameters estimated by the parameter estimation unit 14. Therefore, according to this embodiment, it is possible to improve the estimation accuracy of the interference probability between the analysis model Ma11 of the flexible object and the analysis model Ma21 of the evaluation target. Therefore, according to this embodiment, it is not necessary to set an excessive safety factor for the robot's operation in order to ensure the safety of the robot that manipulates a flexible object. Therefore, it is possible to eliminate constraints on the robot's operation and the schedule of the operation, and therefore it is possible to plan a more efficient operation or schedule of the operation according to this embodiment.

In addition, in the above-described embodiment, the interference assessment unit 15 calculates the probability of interference by performing a physical simulation of a case where a predetermined operation is performed on a flexible object using multiple model parameters obtained by generating random numbers based on the probability distribution of the model parameters. Therefore, according to this embodiment, it is possible to more accurately calculate the probability that a flexible object will interfere with the evaluation target.

In addition, in the above-described embodiment, the interference assessment unit 15 assesses that the flexible object analytical model Ma11 and the evaluation target analytical model Ma21 are interfering with each other when the area occupied by the evaluation target analytical model Ma21 in the virtual space in which the physical simulation is performed, or the interference detection area set in the evaluation target analytical model Ma21 overlaps with the area occupied by the flexible object analytical model Ma11. Therefore, in this embodiment, the state in which actual interference occurs can be reproduced more accurately in the physical simulation, thereby improving the estimation accuracy of the probability of interference between the flexible object and the evaluation target.

In addition, in the above-described embodiment, the simulation unit 13 performs a physical simulation that takes into account the contact force between the analytical model of the flexible object and the analytical model of the evaluation target. Then, the interference evaluation unit 15 evaluates the interference state between the analytical model of the flexible object and the analytical model of the evaluation target based on the magnitude of the contact force in the physical simulation. Therefore, in this embodiment, the state in which interference occurs can be evaluated in a manner close to reality, thereby improving the estimation accuracy of the probability of interference between the flexible object and the evaluation target.

In addition, in the above-described embodiment, the contact force is represented by the stretch of the analytical model of the flexible object or the magnitude of the stress generated in the analytical model of the flexible object, and the interference evaluation unit 15 evaluates that the analytical model of the flexible object and the analytical model of the evaluation object interfere with each other when the contact force exceeds a threshold value previously set for the contact force. Therefore, in this embodiment, the probability of interference between the flexible object and the evaluation object is calculated taking into account situations that are expected to occur when the robot actually operates the flexible object, so that the estimation accuracy of the probability of interference between the flexible object and the evaluation object can be improved.

<Modification>
It should be noted that the present invention is not limited to the above-described embodiment, and various other applications and modifications are possible without departing from the gist of the present invention as set forth in the claims.

[Modification 1]
In the first modification of the embodiment of the present invention, the interference evaluation unit 15 determines that a flexible object and the evaluation object have come into contact with each other when the analytical model of the flexible object comes into contact with an interference detection region provided on the evaluation object.

Figure 8 is a diagram showing an example of an evaluation of interference between a flexible object and an evaluation target by the interference evaluation unit 15 according to the first modified example. As shown in Figure 8, in the virtual space Vs in which the simulation is performed, an analytical model Ma11 of the flexible object, an analytical model Ma21 of the evaluation target, an analytical model Ma31 of the operation unit, and an analytical model Ma32 of the boundary point are provided.

In the example shown in FIG. 8, an interference detection region 211 (a region indicated by diagonal lines slanting downward to the right in the figure) is provided on a portion of the surface of the analytical model Ma21 to be evaluated. In variant example 1, the interference evaluation unit 15 evaluates that interference exists when the analytical model Ma11, which is a flexible object, comes into contact with the interference detection region 211.

According to the first modification, it is possible to appropriately determine whether or not a flexible object interferes with a specific surface of the evaluation target. Note that, in the first modification, an example is given in which the interference detection region is provided on the side surface of the analytical model Ma21 of the evaluation target, but the present invention is not limited to this. The interference detection region may be, for example, a surface (x-y plane) that crosses the evaluation target, or may be another surface.

[Modification 2]
In the second modification, the interference evaluation unit 15 evaluates the interference between the flexible object and the evaluation target in consideration of the contact force when the flexible object comes into contact with the evaluation target. Specifically, the interference evaluation unit 15 performs a physical simulation after adding a contact force, such as a normal force or a frictional force, that the flexible object receives from the evaluation target to the external force in the governing equation for the flexible object, and evaluates the interference between the flexible object and the evaluation target based on the result of the physical simulation.

Figure 9 is a diagram showing an example of an evaluation of interference between a flexible object and an evaluation target by the interference evaluation unit 15 according to the second modified example. In Figure 9 as well, an analytical model Ma11 of the flexible object, an analytical model Ma21 of the evaluation target, an analytical model Ma31 of the operation unit, and an analytical model Ma32 of the boundary point are provided in the virtual space Vs in which the simulation is performed.

In the example shown in FIG. 9, the interference assessment unit 15 sets a threshold value for the elongation of the flexible object analysis model Ma11 or the stress generated in the flexible object analysis model Ma11, and evaluates that there is interference when the measured value obtained by the physical simulation exceeds the threshold value. In other words, in the second modification, the interference assessment unit 15 evaluates interference taking into account not only the information on the presence or absence of contact, but also the information on the contact force at the time of contact.

Therefore, even when the analytical model Ma31 of the operating unit applies a predetermined external force to the analytical model Ma11 of the flexible object, the state (strength, etc.) of the catch of the analytical model Ma11 of the flexible object on the analytical model Ma21 being evaluated will differ depending on the values of the model parameters.

Even in the second modification shown in FIG. 9, the interference assessment unit 15 can calculate the interference probability using the above-mentioned formula (2) etc.

According to the second modification, a judgment is made taking into account the contact force between the flexible object and the evaluation target, so the interference evaluation unit 15 can more precisely evaluate the interference between the flexible object and the evaluation target. The simulation device and simulation method according to the second modification are particularly effective in situations where a task is routed by the action of hooking a flexible object onto the evaluation target, or when it is desired to consider whether or not a task is restricted when a hook occurs.

In the above-described embodiment and modified examples 1 and 2, the interference assessment unit 15 assesses the interference between the flexible object and the evaluation target based on whether the flexible object is caught on the evaluation target, but the present invention is not limited to this. For example, the interference assessment unit 15 may assess the interference between the flexible object and the evaluation target based on whether the flexible object is wrapped around the evaluation target. Furthermore, the criteria for determining interference and the method of displaying the determination results are not limited to the methods shown in the above-described embodiment and modified examples 1 and 2, and various methods may be adopted.

Second Embodiment
Next, a simulation apparatus and a simulation method according to a second embodiment of the present invention will be described with reference to Fig. 10 to Fig. 14. First, the configuration of the simulation apparatus according to the second embodiment will be described with reference to Fig. 10. Fig. 10 is a diagram showing an example of the configuration of a simulation apparatus 10A according to the second embodiment.

The simulation device 10A shown in FIG. 10 differs from the simulation device 10 according to the first embodiment shown in FIG. 1 in that the simulation device 10A according to this embodiment includes a planning unit 18. The planning unit 18 is a block that plans the robot's operation or the robot's operation schedule based on the evaluation result of the interference between the flexible object and the evaluation target by the interference evaluation unit 15. Each block other than the planning unit 18 is similar to the simulation device 10 shown in FIG. 1, so a description of these will be omitted here.

[Example 1]
Fig. 11 is a diagram showing an example of a motion plan of a robot according to Example 1 of the second embodiment. Fig. 11 shows a real environment, in which a boundary point 21 fixed to a wall or the like and an operation unit 22 grasped by a robot or the like (not shown) are connected by a flexible object 23. In addition, an evaluation object 24 is also provided in the real environment shown in Fig. 11, in which interference with the flexible object 23 is evaluated.

In the motion plan according to Example 1, the robot moves the operation unit 22 along the path R shown by a bold dashed line from an initial position P0 shown by a dashed circle to a target position Pt shown by a solid circle without causing the flexible object 23 to interfere with the evaluation target 24. The method of determining interference between the flexible object 23 and the evaluation target 24 by the interference evaluation unit 15 may be, for example, the method shown in Modification 1 (see FIG. 8). That is, the interference evaluation unit 15 evaluates that there is interference when the analytical model of the flexible object comes into contact with an interference determination region provided in the analytical model of the evaluation target 24.

In this embodiment, the interference evaluation unit 15 performs a physical simulation in two patterns: moving the operation unit 22 at a constant speed, and slowing down the moving speed of the operation unit 22 at a position (intermediate position) near the evaluation target.

Figure 12 shows an example of the results of a physical simulation when the operating unit 22 is moved at a constant speed. Figure 12A shows a state in which the analytical model Ma32 of the operating unit has moved from the initial position P0 via the intermediate position Pv to the target position Pt. Figure 12A also shows that in two of the physical simulations performed multiple times, the analytical model Ma11 of the flexible object was wrapped around the analytical model Ma21 of the evaluation target, and that the probability of interference between the flexible object and the evaluation target was 29%.

Figure 12B shows a graph of an example of the transition over time of the movement speed of the analytical model Ma32 of the operating unit. The vertical axis of Figure 12B shows the movement speed (m/s) of the operating unit 22, and the horizontal axis shows the position of the analytical model Ma32 of the operating unit in the virtual space Vs. Figure 12B shows that the movement speed of the analytical model Ma32 of the operating unit is constant (1.0 [m/s]) from position P1, which is a predetermined time after departure from the initial position P0, to position P2 just before the destination position Pt.

Figure 13 shows an example of the results of a physical simulation in which the moving speed of the operating unit 22 is decelerated at a position (intermediate position) near the evaluation target. Figure 13A shows a state in which the analytical model Ma32 of the operating unit has moved from the initial position P0 via the intermediate position Pv to the target position Pt. Figure 13A also shows that in one of the physical simulations performed multiple times, the analytical model Ma11 of the flexible object is wrapped around the analytical model Ma21 of the evaluation target, and that the probability of interference between the flexible object and the evaluation target is 14%.

Figure 13B shows a graph of an example of the transition over time of the moving speed of the analytical model Ma32 of the operating unit. The values indicated by the vertical and horizontal axes of Figure 13B are the same as those in Figure 12B. Figure 13B shows that the moving speed of the analytical model Ma32 of the operating unit decelerates between position P11, which is slightly before the via position Pv, and position P12, which is slightly after the via position Pv.

In the example shown in FIG. 13A where the moving speed of the operating unit 22 is slowed midway, the probability of interference between the flexible object and the evaluation target is 14%, which is lower than the interference probability of 29% when the operating unit 22 is moved at a constant speed as shown in FIG. 12A. This is because by slowing down the moving speed of the operating unit 22 midway, the centrifugal inertia force generated in the flexible object analysis model Ma11 when the operating unit 22 passes around the via position Pv where the speed is slowed down is reduced, and the probability that the flexible object analysis model Ma11 will interfere with the evaluation target analysis model Ma21 is correspondingly lower.

On the other hand, in the example shown in Figures 13A and 13B where the moving speed of the operation unit 22 is decelerated midway, the time it takes to reach the target position Pt is longer by the amount of the deceleration than in the example shown in Figures 12A and 12B where the operation unit 22 is moved at a constant speed.

When interference occurs between a flexible object and the evaluation target, if the time from detection of the interference to recovery from the interference state is "td (min)", the expected value "tm (min)" of the time until the analytical model Ma32 of the operation unit reaches the target position Pt is expressed by the following formula (3).

tm=(1-α)tn+α(tn+td)=tn+αtd…Equation (3)

In the above formula (3), "α" indicates the interference probability, and "tn" is the time to perform (execute) the movement when no interference occurs. The planning unit 18 (see FIG. 10) can optimize the movement speed of the operation unit 22 by repeatedly correcting the movement speed of the operation unit 22 so that tm, which is the expected value of the execution time tn, is minimized.

The method of planning the robot's motion by the planner 18 is not limited to the methods exemplified in Figs. 12 and 13. For example, the planner 18 may probabilistically evaluate the reward used in reinforcement learning based on the interference probability value calculated by the interference evaluator 15, and plan the robot's motion based on the result of the evaluation.

When the safety of the robot's operation is important, the planning unit 18 may set a threshold for the probability of interference and plan the robot's operation so that the probability of interference is equal to or less than the threshold. Furthermore, when the time until interference is detected cannot be uniquely determined, the planning unit 18 may provide the delay time td caused by interference as a probability distribution.

[Example 2]
In Example 2, the planner 18 plans an operation schedule for the robot based on the interference probability calculated by the interference assessor 15. Specifically, the planner 18 creates a schedule for the robot by combining a plurality of operations such as those shown in Example 1 of the second embodiment described above, and assigning a robot to each of the plans.

At this time, the planning unit 18 optimizes the schedule by creating the schedule using a search algorithm such as Dijkstra's algorithm. Note that the method for optimizing the schedule is not limited to Dijkstra's algorithm, and A* algorithms, genetic algorithms, deep learning, etc. can also be used.

In Dijkstra's algorithm, the cost of the schedule set for each node is calculated according to an evaluation function, and the optimal schedule is determined by searching for nodes with good costs. In this embodiment, the execution time of the schedule is assigned to the evaluation function, but instead of the execution time of the schedule, the power consumption of the robot, etc. may also be assigned.

When the execution time of an operation assigned to a robot varies probabilistically, the planning unit 18 constructs a schedule for each probabilistically varying execution time and calculates the expected cost Cost according to the following formula (4).

Cost=tsm=(1-α)tsn+αtsa...Formula (4)

In the above formula (4), "α" indicates the interference probability, "tsm" indicates the expected execution time of the schedule, "tsn" indicates the execution time of the schedule when no interference occurs, and "tsa" indicates the execution time of the schedule when interference occurs.

Figure 14 is a diagram showing an example of the robot motion plan formulated by the planner 18 at node n (n is a natural number equal to or greater than 1). Figure 14 shows an example of the robot motion plan in the case where the interference probability when the robot 102 executes "motion E" is 10% and the recovery time from the interference state is 0.5 min. In addition, in the example shown in Figure 14, it is assumed that there is a constraint that "motion C" and "motion E" cannot be executed at the same time.

Figure 14A shows an example of a robot motion plan in the case where no interference occurs between the flexible object and the evaluation target when the robot 102 performs "motion E," and Figure 14B shows an example of a robot motion plan in the case where interference occurs between the flexible object and the evaluation target when the robot 102 performs "motion E."

Figure 14A shows that the expected value (tsm) of the execution time of the operation by the robot 101 is 5.5 [min], and Figure 14B shows that the expected value (tsa) of the execution time of the operation by the robot 101 is 6 [min]. Therefore, when the expected values tsm and Tsa are substituted into the above formula (4), the cost of node n is calculated to be 5.55 [min]. The planning unit 18 can determine the optimal schedule that minimizes the evaluation function by searching for the node with the smallest cost.

The planning unit 18 may also give the delay time caused by interference using a probability distribution, obtain schedules that change based on the probability distribution, and calculate the expected value tsm based on the expected values of those costs.

In the second embodiment described above, the planning unit 18 plans the robot's motion and/or the robot's motion schedule based on the probability of interference between the flexible object analysis model Ma11 and the evaluation target analysis model Ma21. Therefore, the planning unit 18 can plan the robot's motion and/or the robot's motion schedule that can ensure the safety of the robot's manipulation of flexibility based on the probability of interference between the flexible object analysis model Ma11 and the evaluation target analysis model Ma21. Therefore, according to this embodiment, it is possible to prevent excessive setting of the safety factor of the robot's motion and/or the motion schedule, and it is also possible to improve the efficiency of the robot's motion and/or the motion schedule.

The planning unit 18 can plan not only the robot's motion and/or motion schedule, but also the placement position or placement method of the flexible object, etc., based on the probability of interference between the flexible object analysis model Ma11 and the evaluation target analysis model Ma21. Specifically, the planning unit 18 can take measures such as fixing positions where the flexible object is likely to get caught to boundary points of pillars, etc.

The above-mentioned embodiment examples are provided to explain the device and system configurations in detail and specifically in order to clearly explain the present invention, and are not necessarily limited to those that include all of the configurations described.

In addition, the control lines or information lines indicated by arrows in Figure 1 are those considered necessary for explanation, and do not necessarily show all control lines or information lines in the product. In reality, it can be assumed that almost all components are interconnected.

In addition, in this specification, the processing steps describing chronological processing include not only processing that is performed chronologically in the order described, but also processing that is not necessarily performed chronologically but is performed in parallel or individually (for example, parallel processing or processing by objects).

Furthermore, each component of the simulation device according to one embodiment of the present disclosure described above may be implemented in any hardware as long as each piece of hardware can transmit and receive information to and from each other via a network. Furthermore, the processing performed by a certain processing unit may be realized by a single piece of hardware, or may be realized by distributed processing using multiple pieces of hardware.

10, 10A...simulation device, 11...storage unit, 12...analysis model construction unit, 13...simulation unit, 14...parameter estimation unit, 15...interference evaluation unit, 16...display control unit, 17...display unit, 18...planning unit

Claims (5)

a memory unit that stores a measurement result of dynamic displacement of a flexible object in a time direction as a behavior of the flexible object when a predetermined operation is performed on an operation unit connected to an end of the flexible object , the flexible object being an indefinite object whose shape is not easily determined and whose one end is fixed to a boundary point ;
a simulation unit that analyzes a behavior of an analytical model of the flexible object when the operation is performed by a physical simulation;
a parameter estimation unit that modifies model parameters of the analytical model of the flexible object so that the behavior of the flexible object stored in the storage unit coincides with the behavior of the analytical model of the flexible object analyzed by the simulation unit, and estimates a probability distribution of the modified model parameters ;
an interference evaluation unit that calculates an interference probability between an analytical model of the flexible object and an analytical model to be evaluated for interference with the flexible object by analyzing a behavior of the flexible object with the simulation unit;
The parameter estimation unit estimates a probability distribution of the model parameters using a data assimilation technique;
the simulation unit sets an external force according to a time history in the dynamic displacement of the flexible object read from the storage unit in an analysis model of the operation unit, and analyzes the behavior of the flexible object based on a probability distribution of the model parameters;
the interference assessment unit calculates the interference probability by performing a physical simulation in a case where a predetermined operation is performed on the flexible object, using a plurality of model parameters obtained by generating random numbers based on a probability distribution of the model parameters; and
when an area occupied by the analytical model of the evaluation target in a virtual space in which the physical simulation is performed, or an interference detection area set in the analytical model of the evaluation target, overlaps with an area occupied by the analytical model of the flexible object, the analytical model of the flexible object is evaluated to interfere with the analytical model of the evaluation target;
Simulation device. the simulation unit performs a physical simulation taking into account a contact force between an analytical model of the flexible object and an analytical model of the evaluation target;
The simulation device according to claim 1 , wherein the interference evaluation unit evaluates an interference state between the analysis model of the flexible object and the analysis model to be evaluated, based on a magnitude of the contact force in the physical simulation. the contact force is represented by an elongation of the analytical model of the flexible object or a magnitude of a stress generated in the analytical model of the flexible object;
When the contact force exceeds a threshold value that is set in advance for the contact force, the interference assessment unit
The simulation device according to claim 2 , wherein the analysis model of the flexible object and the analysis model of the evaluation target are evaluated as interfering with each other. The simulation device according to claim 1 , further comprising a planning unit that plans an operation of a robot that manipulates the flexible object and/or a schedule for the operation of the robot, based on the interference probability acquired by the interference evaluation unit. A simulation method using a simulation device having a memory unit that stores a measurement result of dynamic displacement of a flexible object in a time direction as a behavior of the flexible object when a predetermined operation is performed on an operation unit connected to an end of the flexible object , the flexible object being an indefinite object having one end fixed to a boundary point and a shape that is difficult to determine, the simulation method comprising:
a simulation unit analyzing, by physical simulation, a behavior of an analytical model of the flexible object when the operation is performed;
a parameter estimation unit correcting model parameters of the analytical model so that the behavior of the flexible object stored in a storage unit coincides with the behavior of the analyzed analytical model of the flexible object, and estimating a probability distribution of the corrected model parameters ;
an interference evaluation unit includes a procedure in which the simulation unit analyzes the behavior of the flexible object, thereby calculating an interference probability between an analytical model of the flexible object and an analytical model to be evaluated for interference with the flexible object;
The parameter estimation unit estimates a probability distribution of the model parameters using a data assimilation technique;
the simulation unit sets an external force according to a time history in the dynamic displacement of the flexible object read from the storage unit in an analysis model of the operation unit, and analyzes the behavior of the flexible object based on a probability distribution of the model parameters;
the interference assessment unit calculates the interference probability by performing a physical simulation in a case where a predetermined operation is performed on the flexible object, using a plurality of model parameters obtained by generating random numbers based on a probability distribution of the model parameters;
when an area occupied by the analytical model of the evaluation target in a virtual space in which the physical simulation is performed, or an interference detection area set in the analytical model of the evaluation target, overlaps with an area occupied by the analytical model of the flexible object, the analytical model of the flexible object is evaluated to interfere with the analytical model of the evaluation target;
Simulation method.

Images