JP4121587B2 - Robot simulation device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a robot simulation apparatus used for behavior control of an industrial or entertainment autonomous robot or a virtual robot existing on a computer.
[0002]
[Prior art]
Conventionally, many robots are assembled in a predetermined shape by combining a plurality of constituent units in a predetermined state with a predetermined correlation, and by controlling the driving of each constituent unit, The movement is changed.
[0003]
For example, the robot disclosed in Japanese Patent Application Laid-Open No. 5-245784 can be constructed in a desired shape by combining a plurality of joint modules and a plurality of arm modules. The robot sets a unique number for each joint module, and the control unit recognizes the connection order of each joint module based on the unique number obtained by communication with each joint module. The control program can be rewritten to a suitable one.
[0004]
Also, virtual robots and virtual creatures are moved by computer graphics (CG) on the computer, and the behavior of the virtual robots and virtual creatures is changed according to input information by sensors connected to the computer. Such a computer game is known.
[0005]
Furthermore, when simulating a real robot, a model with the same shape and mass as the real robot is built on a computer, and physical forces in the real world such as gravity are applied to this to calculate the movement. The method of obtaining by is described in many textbooks of robots.
[0006]
[Problems to be solved by the invention]
By the way, since the robot disclosed in Japanese Patent Laid-Open No. 5-245784 is based on the premise of a manipulator, it cannot cope with a configuration in which two or more component units are branched and connected. Also, it does not correspond to control based on input information from various sensors. Therefore, the applicant of the present invention previously described as Japanese Patent Application No. 9-19040, in a robot apparatus composed of a plurality of component units, a first storage means for storing shape information for determining the shape of the component units, and a configuration A second storage means for storing movement information necessary for describing the movement of the unit, a third storage means for storing characteristic information of the electronic component housed in the constituent unit, and a coupling state of each constituent unit. It has been proposed to provide detection means for detecting. The control means in this robot apparatus can automatically recognize the entire structure and the motion characteristics of each component unit based on the detection result by the detection means. In the robot apparatus, the control unit uses the control program used to control each component unit to obtain the first data represented in a predetermined data format that is defined in advance for each function of each electronic component. A conversion program for converting to second data represented in a data format used for each function by each electronic component is stored in each storage unit of each component unit. As a result, each component unit can be designed without depending on the data format predetermined by the control program.
[0007]
In this way, the shape and function information is stored in the robot parts, and by connecting them with a signal line connection method (serial bus) having a mechanism capable of knowing the order of connection, the CPU that controls the robot has the shape of the robot. And you can know the types of sensors and actuators and where they are attached.
[0008]
According to the technology disclosed in Japanese Patent Application No. 9-19040, the information held by the robot parts named virtual robots and the software objects managing the connection order are automatically determined from the information held by the serial bus and robot parts. The robot can be constructed and controlled to control the actual robot.
[0009]
SUMMARY OF THE INVENTION An object of the present invention is to configure a common robot component for a real robot and a virtual robot so that a program for controlling the same can be used in common.
[0010]
[Means for Solving the Problems]
The present invention constructs the virtual robot by giving a combination instruction on the program to software having the same configuration as the information or program stored in the actual robot part, and the program controlling the virtual robot is assigned to the virtual robot. It is characterized by controlling the virtual robot by acting.
[0011]
The robot simulation apparatus according to the present invention is stored for each of the above components for obtaining the center of gravity, mass, rotation axis, and inertia matrix at the center of gravity in the coordinate system set for each of a plurality of components constituting the robot. Based on the information, in a coordinate system set for each of a plurality of components constituting the robot, a first calculation means for obtaining an inertia matrix at the center of gravity, mass, rotation axis, and center of gravity, and acquisition order of the components And when the robot is a virtual robot, the first calculating means includes: an acquiring means for performing the coordinate transformation between the components connected using the acquired joining order. Accordingly, in the coordinate system set to each virtual component constituting the virtual robot, the center of gravity, mass, rotary shaft, the inertia matrix at the center of gravity Performed Mel operation, performs a coordinate transformation between each virtual component by the second calculating means, if said robot is a real robot, by the first calculating means, constituting the actual robot each in each of the components set coordinate system, performs centroid, mass, rotary shaft, the calculation for obtaining the inertia matrix at the center of gravity, and performs coordinate transformation between the respective components by the second arithmetic means .
[0012]
In the robot simulation apparatus according to the present invention, when the robot is a virtual robot, the coordinate system set for each virtual component stored in the virtual component constituting the virtual robot is stored. The virtual robot is configured by the first calculation means using the center of gravity, the mass, the rotation axis, the information for obtaining the inertia matrix at the center of gravity and the information for performing coordinate conversion between the connected virtual components. In the coordinate system set for each virtual component, the calculation is performed to obtain the center of gravity, mass, rotation axis, inertia matrix at the center of gravity, and the second calculation means performs coordinate conversion between the virtual components, Simulate the movement of a virtual robot .
[0014]
When the robot is a real robot, the robot simulation apparatus according to the present invention has a center of gravity and a mass in a coordinate system stored in each component constituting the real robot and set in each component. , The rotation axis, the information for obtaining the inertia matrix at the center of gravity, and the information for performing the coordinate conversion between the connected components , the first computing means is used for each component constituting the real robot, respectively. In the set coordinate system, a calculation for obtaining the center of gravity, mass, rotation axis, and inertia matrix at the center of gravity is performed, and the second arithmetic unit performs coordinate conversion between the above-described components to simulate the movement of the real robot.
[0015]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[0016]
First, a schematic configuration of an actual robot is shown in FIG. The real robot shown in FIG. 1 includes an agent model unit 10 and a CPC (Configurable Physical Component) unit 20. The agent model unit 10 includes a CPU 11, a RAM 12, a ROM 13, a serial bus host controller 14, and a general CPU peripheral device 15. Master input / output is performed via the serial bus host controller 14. .
[0017]
In addition, the CPC unit 20 is coupled as a robot component while CPC components 21A, 21B, 21C, 21D, 21E,... Which are physical components are coupled or branched in a tree structure. The robot component, that is, the CPC component includes a HUB 22 that branches the signal line of the coupled serial bus, a signal processing unit 23, and a memory 24. The signal processor 23 for processing the serial bus signal has a controller on the serial bus device side, and performs clock synchronization, error detection, data packet retransmission request, address management, and the like. In addition, the memory 24 is used to control the part data by transferring it to the shape data of the robot part, the physical data necessary for establishing the equation of motion, the function information of the part, or the agent model unit. Programs and the like are stored, and in response to a request from the agent model unit 10, these data and programs are transferred from the serial bus or a control command is input.
[0018]
Next, FIG. 2 shows the software configuration of the actual robot divided into functional components. That is, the agent model unit 10 of the real robot includes a serial bus host controller (hardware and software) 10A that controls the serial bus, and a virtual robot (software) 10B that manages information and connection of each CPC component of the CPC unit. The blueprint robot (software) 10C that defines the meaning of each CPC part in the robot according to the Robot Part Semantics data given by the user, and the behavior model and behavior parameters to make the blueprint robot 10C function as an autonomous robot It comprises an agent controller (software) 10D that performs sensor processing, behavior control, etc.
[0019]
The CPC unit 20 is functionally composed of a serial bus device controller 20A and a robot function unit 20B. The serial bus device controller 20A processes signals input / output from the serial bus, and functionally exchanges data with the serial bus host controller 10A of the agent model unit 10 using a protocol. The robot function unit 20B operates in accordance with data from the serial bus device controller 20A or an instruction from the agent model unit 10, and functionally has a protocol with the virtual robot 10B of the agent model unit 10 to exchange data. Do.
[0020]
In the real robot having such a configuration, serial data input / output processing is performed on the agent model unit 10 side by the hardware of the serial bus host controller 10A and the software for controlling the hardware. That is, at the time of data output, the serial bus host controller 10A adds a synchronization sync pattern, header information, data, error detection, and correction code to the data, and transmits the data after performing modulation such as NRZ. After the data is sent, an acknowledge signal or the like from the destination is received, and after confirming that the data has been acquired, the next data processing is started. At the time of data input, the serial bus host controller 10A performs clock synchronization, sync pattern detection, header information processing, data acquisition, error detection, correction, and the like of the transmitted signal, and sends an acknowledge signal. If there is an error in the data, a retransmission request signal is output. In this way, the serial bus host controller 10A performs data input / output via the serial bus.
[0021]
Further, the connecting order of the CPC parts 21A, 21B,... Of the CPC unit 20 can be known using the fact that the CPC parts 21A, 21B,. That is, the serial bus host controller 10A can know the connecting order of the CPC components 21A, 21B,... By associating the address for passing information to each CPC component with the branch order by the serial bus host controller 10A. On the other hand, the branch coupling portion of the CPC component to the next component is stored as information in the CPC component, and using this information, the shape of the CPC component, the position to be coupled next, and the coupling are coupled there. The serial bus host controller 10A can know the shape of the component. Therefore, when the serial bus host controller 10A passes the data to the virtual robot 10B, the virtual robot 10B is a robot that is in what shape, has what function, and where it is. Information for recognizing the existence can be automatically acquired.
[0022]
That is, the movement of the actual robot, the coordinate system set in a plurality of components that constitute the actual robot, the center of gravity, mass, rotary shaft, obtains the inertia matrix at the center of gravity, the coordinate transformation between linked components By doing so, a simulation can be performed.
[0023]
Therefore, in the actual robot having the configuration shown in FIGS. 1 and 2, the CPC parts 21A, 21B,... Which are physical components are combined as a robot part while being combined or branched in a tree structure. The CPC unit 20 includes the CPC unit 20 and the agent model unit 10 which acquires the information of the CPC components 21A, 21B...・ The drive can be controlled.
[0024]
Next, a virtual robot to which the present invention is applied will be described.
[0025]
The virtual robot to which the present invention is applied includes a virtual CPC unit that combines virtual robot parts that are virtual components instead of physical components in the real robot. Drive control of each virtual robot part is performed. That is, the real robot part is handled as software or virtual robot part that holds only the internal information, and the virtual robot is controlled by the program controlling the real robot.
[0026]
Here, the virtual robot component is software or data displayed using computer graphics (CG) by a workstation or the like, for example.
[0027]
The virtual robot shown in FIG. 3 has a configuration in which the CPC unit 20 is replaced with a virtual CPC unit 120 without changing the agent model unit 10 in the real robot shown in FIGS.
[0028]
That is, the agent model unit 110 of the virtual robot includes a serial bus controller (hardware and software) 110A, a virtual robot (software) 110B, a blueprint robot (software) 110C, and an agent controller (software) 110D. The hardware configuration and the software configuration are the same as the agent model unit 10 in the real robot described above.
[0029]
The virtual CPC unit 120 is configured only by the information of the CPC unit in the above-described real robot, and connects not the physical CPC component but the virtual robot functional component including the data or program existing on the computer. There is a virtual CPC manager 120B made of software.
[0030]
In the virtual CPC unit 120, the agent model unit 110 communicates with the virtual CPC manager 120B through the same serial bus as the serial bus in the real robot, and the virtual CPC manager 120B interprets the serial bus protocol. The virtual serial bus device controller 120A, which reacts in response, is in charge of communication.
[0031]
Further, a viewer 120C (software) for showing the movement of the robot to the user is further present on the virtual CPC unit 120 side. In the above-described real robot, the robot or CPC component receives a force such as gravity in the real physical world, but the virtual robot performs a process such as gravity using software. That is, the virtual world manager 120D emulates a virtual world in which a virtual robot part exists, and the virtual world is drawn on the display by the viewer 120C.
[0032]
In order to draw a virtual world on the display, a general CG technique can be used. For example, the received light level at the observation point may be displayed on the display for the reflected light of the object defining the surface reflection model assuming the direction of the light source. Alternatively, as a simpler drawing method, a two-dimensional projection may be drawn assuming a plane perpendicular to the viewpoint direction of the three-dimensional object.
[0033]
In this way, the virtual robot is constructed by giving a combination instruction on the program to software having the same configuration as the information or program stored in the actual robot part, and the program that has controlled the virtual robot is transferred to the virtual robot. The virtual robot can be controlled by acting.
[0034]
The movement of the virtual robot is determined by calculating the center of gravity, mass, rotation axis, and inertia matrix at the center of gravity in the coordinate system set for each of the virtual components constituting the virtual robot. A simulation can be performed by performing the conversion.
[0035]
In addition, information for obtaining the center of gravity, mass, rotation axis, inertia matrix at the center of gravity and information for performing coordinate conversion between connected virtual components is stored in the coordinate system attached to each virtual component. By doing so, it is possible to perform the simulation by freely removing or adding.
[0036]
Therefore, the virtual robot is constructed by giving a combination instruction on the program to software having the same configuration as the information or program stored in the actual robot part, and the program controlling this is applied to this virtual robot. Thus, the virtual robot can be controlled, and the virtual robot can be controlled by the virtual robot control program obtained by the virtual world manager simulating the virtual robot part or the virtual robot.
[0037]
Note that the virtual world manager's method of simulating virtual robot parts or virtual robots is as follows. There are ways to solve it.
[0038]
In the textbook of robot control theory, the theory of robot control is explained as a problem of reverse kinematics and forward kinematics of robots.
[0039]
Inverse kinematics is the robot angle θ (t) [p × 1]
Angle θ '(t) [p × 1]
Angle θ "(t) [p × 1]
Are given, the general equation of motion J (θ) θ ″ + C (θ ′, θ) + Dθ ′ + P (θ) + E (θ ′, θ) = τ
This is a problem of obtaining the torque on the right side by directly substituting for the left side. here,
J (θ) θ "is the inertia term of p × 1,
C (θ ', θ) is the term for the centrifugal force of p × 1, the Coriolis force,
Dθ ′ is the p × 1 viscous friction coefficient P (θ) is the p × 1 gravity term E (θ ′, θ) is the p × 1 nonlinear friction term,
τ (t) is p × 1 input torque.
[0040]
By solving such an inverse kinematics problem, for example, when a target angle, angular velocity, and angular acceleration are given, a torque necessary to perform such a motion can be obtained.
[0041]
As a method of deriving the equation of motion in the inverse kinematics problem, paying attention to various energy functions of the robot, substituting them into the Lagrangian equation, the balance of force and moment for each link of the robot, action, reaction In particular, there is known a method for obtaining the motion equation of the entire robot by using the input equation of motion and Euler's equation of motion.
[0042]
The forward dynamics problem is the kind of motion that the robot performs when given the initial state of the robot [θ (0), θ (0), θ (0)] and the input torque τ (t). The equation of motion J (θ) θ "+ C (θ ', θ) + Dθ' + P (θ) + E (θ ', θ) = τ
It is a problem to ask based on. That is, the forward dynamics problem is a problem of solving a differential equation, and the above equation is expressed as θ ″ = J (θ) ⁻¹ [−C (θ ′, θ) −Dθ′−P (θ) −E (θ ', θ) + τ]
After deformation, an appropriate initial condition θ (0) = θ ₀
θ '(0) = θ ₀ '
Can be solved by performing numerical integration using the Euler method or the Rungetac method.
[0043]
For example, the forward dynamics problem can be solved by the procedure shown in the flowchart of FIG.
[0044]
In step S1, initial setting is performed at t = 1 and i = 1.
[0045]
In step S2,
h (t) _i = INV [θ (t), 0, e _i , 0]
Perform the operation.
[0046]
In the next step S3, it is determined whether i = p. If i = p is not satisfied, i = i + 1 is set in step S4, the process returns to step S2, and the process of calculating the next h (t) _i is repeated until i = p. Calculate the column. If i = p, the process proceeds to step S5.
[0047]
In step S5,
b (t) = INV [θ (t), θ (t) ′, 0, g]
Calculate b (t) by
[0048]
In the next step S6,
θ "(t) = H (t) ^-1 [τ (t) -b (t)]
To calculate θ ″ (t).
[0049]
In the next step S7,
θ ′ (t + Δt) = θ ′ (t) + θ ″ (t) Δt
θ (t + Δt) = θ (t) + θ '(t) Δt + θ "(t) Δt 2/2
Calculate
[0050]
In the next step S8,
t = t + Δt
And
[0051]
In the next step S9, it is determined whether or not t> t _end . If t> t _{end is} not satisfied, the process returns to step S2, and the processes from step S2 to step S8 are repeated. If t> t _end , the process ends.
[0052]
【The invention's effect】
As described above, according to the present invention, it is possible to use a common robot component configuration for a real robot and a virtual robot and to use a program for controlling the same in common.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a schematic configuration of a real robot to which the present invention is applied.
FIG. 2 is a diagram showing a software configuration of the real robot divided into functional components.
FIG. 3 is a diagram showing a software configuration of a virtual robot to which the present invention is applied divided into functional components.
FIG. 4 is a flowchart showing a general solution to a forward dynamics problem.
[Explanation of symbols]
10, 110 agent model part, 10A, 110A serial bus controller, 10B, 110B virtual robot, 10C, 110C blueprint robot, 10D, 110D agent controller, 11 CPU, 12 RAM 13 ROM, 14 host controller 14, 20 CPC part, 20A Serial bus device controller, 20B Robot functional unit 21A, 21B, 21C, 21D, 21E ... CPC parts, 22 HUB, 23 signal processing unit, 24 memory, 120A virtual serial bus device controller, 120B virtual CPC manager, 120C viewer, 120D Virtual World Manager

Claims (3)

The robot is configured based on information stored for each component for obtaining the center of gravity, mass, rotation axis, and inertia matrix at the center of gravity in a coordinate system set for each of the plurality of components constituting the robot. In a coordinate system set for each of a plurality of component parts,
An acquisition means for acquiring a joining order of the components ;
A second computing means for performing coordinate transformation between components connected using the obtained coupling order,
When the robot is a virtual robot, the first calculation means obtains the center of gravity, mass, rotation axis, and inertia matrix at the center of gravity in the coordinate system set for each virtual component constituting the virtual robot. Perform computation, perform coordinate transformation between the virtual components by the second computing means,
When the robot is a real robot, the first calculation means calculates the center of gravity, mass, rotation axis, and inertia matrix at the center of gravity in the coordinate system set for each component constituting the real robot. And performing coordinate transformation between the component parts by the second computing means . When the robot is a virtual robot, the center of gravity, the mass, the rotation axis, and the rotation axis in the coordinate system respectively set for each virtual component stored in the virtual component constituting the virtual robot are stored . Using information for determining the inertia matrix at the center of gravity and information for performing coordinate transformation between connected virtual components ,
In the coordinate system set for each virtual component constituting the virtual robot by the first calculation means, calculation is performed to obtain the center of gravity, mass, rotation axis, inertia matrix at the center of gravity, and the second calculation means Perform coordinate transformation between each virtual component above,
The robot simulation apparatus according to claim 1 , wherein the movement of the virtual robot is simulated . When the robot is a real robot , the center of gravity, the mass, the rotation axis, and the inertia matrix at the center of gravity are stored in the coordinate system stored in each component constituting the real robot. Using the information to obtain and the information to perform coordinate transformation between connected components ,
In the coordinate system set for each component constituting the real robot by the first calculation means, calculation is performed to obtain the center of gravity, mass, rotation axis, and inertia matrix at the center of gravity, and the second calculation means performs the above-described calculation. Perform coordinate transformation between each component,
The robot simulation apparatus according to claim 1, wherein the movement of the real robot is simulated.

Images