CN110065070A - A kind of robot adaptive impedance control system based on kinetic model - Google Patents
A kind of robot adaptive impedance control system based on kinetic model Download PDFInfo
- Publication number
- CN110065070A CN110065070A CN201910352004.9A CN201910352004A CN110065070A CN 110065070 A CN110065070 A CN 110065070A CN 201910352004 A CN201910352004 A CN 201910352004A CN 110065070 A CN110065070 A CN 110065070A
- Authority
- CN
- China
- Prior art keywords
- robot
- indicate
- torque
- moment
- controller
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1602—Programme controls characterised by the control system, structure, architecture
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1628—Programme controls characterised by the control loop
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1628—Programme controls characterised by the control loop
- B25J9/1633—Programme controls characterised by the control loop compliant, force, torque control, e.g. combined with position control
Abstract
The invention belongs to industrial robots to control correlative technology field, the robot adaptive impedance control system based on kinetic model that it discloses a kind of, the system includes preprocessing module, impedance controller and adaptive strategy control module, preprocessing module is for constructing Dynamic Models of Robot Manipulators, and the kinetic parameter and friction coefficient of robot are recognized, and then optimize to Dynamic Models of Robot Manipulators；It is also used to the Dynamic Models of Robot Manipulators after optimization being transferred to impedance controller；Impedance controller is used to realize the power of robot end and the Shared control of position according to Dynamic Models of Robot Manipulators, and the location error being calculated, velocity error and external force value are transferred to adaptive strategy control module；Adaptive strategy control module is used to be judged according to the data received and the numerical value itself prestored, and calculates controller parameter, while controller parameter is transferred to impedance controller.Adaptivity of the present invention is preferable, and precision is higher, and flexibility is good.
Description
Technical field
The invention belongs to industrial robots to control correlative technology field, be based on kinetic model more particularly, to one kind
Robot adaptive impedance control system.
Background technique
With the development of science and technology, industrial robot starts to be widely used in the every field such as intelligence manufacture and aerospace.It is right
Robot demand and required precision are also higher and higher, and traditional control method is controlled to the position of joint of robot
System, may cause joint of robot torque transfinite either end contact force it is excessive, cause workpiece damage in handling process, accidentally
In the presence of difference part equipment failure or machining precision it is low the problems such as, it is more serious may make robot end or
Person's connecting rod is impaired.Particularly, to the assembly work of certain working environment or high-precision requirements complicated and changeable, position control is difficult
To reach the requirement of process and assemble precision.
Currently, commonly the control method based on torque mainly includes impedance control and the mixing of power position to industrial robot system
Control.Simplest method is to obtain contact torque by installing force snesor or torque sensor in robot end,
But the method increase the complexity of robot architecture and costs.
Wherein, impedance control is divided into location-based impedance control and the impedance control based on torque again, location-based
Impedance control is coordinated to the position and speed of robot end, and end torque is unable to control；Impedance control based on torque
System is usually used in the case where machine human and environment contact, and in impedance control, in robot end, there are application values when contact force
It is larger, but when torque is excessive will affect PD control device model for contact, reduce control precision.Correspondingly, this field there is
Develop a kind of technical need of the higher robot adaptive impedance control system based on kinetic model of precision.
Summary of the invention
Aiming at the above defects or improvement requirements of the prior art, the present invention provides a kind of machines based on kinetic model
People's adaptive impedance control system, be based on existing impedance control feature, in order to solve the problems, such as control precision it is lower, research and
Devise a kind of precision preferably robot adaptive impedance control system based on kinetic model.The control system is based on
The Dynamic Models of Robot Manipulators established carries out Shared control to robot, does not need mounting torque sensor, can be by dynamic
Mechanical model and current of electric estimate moment of face, and based on ADAPTIVE CONTROL according to location error, velocity error, moment of face
And the value of given stabilization moment of face obtains the controller parameter of new impedance controller, to realize the balance of position and torque
Control, adaptivity is preferable, and precision is higher.
To achieve the above object, according to one aspect of the present invention, a kind of robot based on kinetic model is provided
Adaptive impedance control system, the control system include preprocessing module, impedance controller and adaptive strategy control module,
Wherein:
The preprocessing module is used for according to the Coulomb friction power and viscous force of joint of robot, using Newton―Leibniz formula
Dynamic Models of Robot Manipulators is constructed, and based on the Dynamic Models of Robot Manipulators using the power of particle swarm algorithm identification robot
Parameter and friction coefficient are learned, and then the Dynamic Models of Robot Manipulators is optimized；Meanwhile the preprocessing module is also used
In by optimization after Dynamic Models of Robot Manipulators be transferred to the impedance controller；
The impedance controller is used to realize power and the position of robot end according to the Dynamic Models of Robot Manipulators
Shared control, and location error, velocity error and the external force value that itself is calculated are transferred to the adaptive strategy control
Molding block；
The adaptive strategy control module be used for when robot end is by external force according to the data received and oneself
The numerical value that body prestores is judged, and correspondingly calculates controller parameter according to judging result, while the controller being joined
Number is transferred to the impedance controller, to update the parameter of the impedance controller, realizes the flat of robot location and torque
Weighing apparatus control.
Further, the preprocessing module is based on the Dynamic Models of Robot Manipulators, using Fourier space conduct
Excitation track recognizes the kinetic parameter and friction coefficient of robot.
Further, the theoretical joint moment being calculated based on the Dynamic Models of Robot Manipulators are as follows:
In formula, q,Respectively indicate robot joint angles, angular speed and angular acceleration；Indicate that robot is dynamic
The moment of inertia of mechanical model prediction；Indicate the centrifugal force and section's formula torque of Dynamic Models of Robot Manipulators prediction；
Indicate the gravitational moment that Dynamic Models of Robot Manipulators calculates；Indicate Dynamic Models of Robot Manipulators prediction
Viscous friction torque and Coulomb friction torque, d is viscous friction coefficient, and μ is Coulomb friction coefficient.
Further, the kinetic parameter of the connecting rod i of robot are as follows:
λi=[mi,si,x,si,y,si,z,Ii,xx,Ii,yy,Ii,zz,Ii,xy,Ii,xz,Ii,yz,di,μi]T
In formula, i=1,2 ..., n, n are amount of articulation；miIndicate the quality of connecting rod i；si,xIndicate the mass center of connecting rod i in xiSide
To component；si,yIndicate the mass center of connecting rod i in yiThe component in direction；si,zIndicate the mass center of connecting rod i in ziThe component in direction,
Ii,xx=∫ ∫ ∫V(yi 2+zi 2)ρdυ、Ii,yy=∫ ∫ ∫V(xi 2+zi 2)ρdυ、Ii,zz=∫ ∫ ∫V(xi 2+yi 2) ρ d υ respectively indicates robot
Load is around ending coordinates axis xi,yi,ziMass mement of inertia；Around xiyi,xizi,yiziThe moment of inertia in direction is expressed as Ii,xy
=∫ ∫ ∫Vxiyiρdυ、Ii,xz=∫ ∫ ∫Vzixiρdυ、Ii,yz=∫ ∫ ∫Vyiziρdυ；ρ indicates connecting rod density；υ indicates connecting rod volume；di
For the viscous friction coefficient of joint i；μiFor the Coulomb friction coefficient of joint i.
Further, the impedance controller includes position and rate control module, prediction torque feed-forward module and contact
Torque-feedback module, wherein the position and rate control module are used for calculating robot in the location error of cartesian space
And velocity error, and then calculated according to the location error and velocity error being calculated and eliminate corresponding location error and speed mistake
Difference needs the active force applied to robot end, then the active force is converted to the opplied moment of joint space, and then controls
Robot motion processed is to reduce the location error and velocity error of robot end.
Further, when work, the prediction theoretical joint power of the torque feed-forward module based on dynamics calculation machine people
Square, and the theoretical joint moment being calculated is transferred to the contact torque feedback module；The contact torque feedback module
The practical joint moment of robot is calculated in electric current for the motor according to robot, and is based on the practical joint moment
And the theoretical joint moment carrys out the external force that calculating robot end is subject in cartesian space, is then transferred to the external force
The adaptive strategy control module, while the robot is controlled according to the external force.
Further, the impedance controller realizes joint by the module and carriage transformation matrix and Jacobian matrix of robot
Position, speed and the torque in space to the position of cartesian space, speed and torque conversion, thus realize to robot end
The control of position, speed and torque.
Further, the kinetics equation of robot are as follows:
In formula, J indicates robot Jacobian matrix；Kp、KdRespectively indicate stiffness coefficient matrix and the damping of impedance controller
Coefficient matrix；xd、Respectively indicate the theoretical position and theoretical velocity of robot end；x,Respectively indicate robot end's
Physical location and actual speed；KeIndicate contact force coefficient matrix；xeIt indicates when elastic deformation does not occur for robot end surface
Position.
Further, the adaptive strategy control module starts to act on when robot end has external force, is based on BP
The controller parameter of neural network algorithm calculating robot, and the obtained controller parameter is transferred to the impedance control
Device.
Further, the stationary value of robot end's location error, velocity error, power and torque error and moment of face is made
For the input signal of the BP neural networkF'=(JT)-1τ is robot end's
Actual forces；For the theoretical power of robot end, FεFor the stationary value of robot end's external force；xd、Table respectively
Show the theoretical position and theoretical velocity of robot end；x,Respectively indicate the physical location and actual speed of robot end.
In general, through the invention it is contemplated above technical scheme is compared with the prior art, base provided by the invention
It is mainly had the advantages that in the robot adaptive impedance control system of kinetic model
1. the present invention is based on the modes of kinetic model prediction robot torque to be controlled, it can effectively reduce and follow mistake
Difference improves precision；Meanwhile, it is capable to influence of the impedance controller parameter value to control effect is reduced, it only need to be at robot end
The self adaptive control that controller parameter is carried out when holding by external force, can effectively reduce the computation complexity of robot controller,
Improve rate.
2. the impedance controller realizes joint space by the module and carriage transformation matrix and Jacobian matrix of robot
Position, speed and torque to the position of cartesian space, speed and torque conversion, be not necessarily to mounting torque sensor, cost compared with
It is low.
3. the adaptive strategy control module starts to act on when robot end has external force, it is based on BP neural network
The controller parameter of algorithm calculating robot, and the obtained controller parameter is transferred to the impedance controller, wherein
It, can be according to the environment or job specification that robot end contacts using the stationary value of robot end's moment of face as input parameter
The autonomous setting moment of face of difference stationary values, realize the automatic adjusument of impedance controller parameter.
4. the position and rate control module calculating robot be in the location error and velocity error of cartesian space, into
And it needs according to the corresponding location error of location error and velocity error calculating elimination and velocity error being calculated to machine
The active force that people end applies, then the active force is converted to the opplied moment of joint space, and then controller robot transports
Dynamic location error and velocity error to reduce robot end, improves control precision, with strong applicability, flexibility is preferable.
5. the structure of robot adaptive impedance control system provided by the invention is simple, easy to implement, be conducive to promote
Using.
Detailed description of the invention
Fig. 1 is the robot adaptive impedance control system based on kinetic model that better embodiment of the present invention provides
Impedance controller control block diagram；
Fig. 2 is the Partial controll block diagram of the robot adaptive impedance control system based on kinetic model in Fig. 1.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.As long as in addition, technical characteristic involved in the various embodiments of the present invention described below
Not constituting a conflict with each other can be combined with each other.
Please refer to Fig. 1 and Fig. 2, the robot adaptive impedance control system provided by the invention based on kinetic model,
The control system includes preprocessing module, impedance controller and adaptive strategy control module, and the preprocessing module is being examined
In the case where Coulomb friction power and the viscous force of considering joint of robot, Dynamic Models of Robot Manipulators is established based on Newton―Leibniz formula,
And kinetic parameter and friction coefficient based on the Dynamic Models of Robot Manipulators and particle swarm algorithm identification robot, with excellent
Change the Dynamic Models of Robot Manipulators.Meanwhile the preprocessing module also the Dynamic Models of Robot Manipulators is transferred to it is described
Impedance controller.
The preprocessing module be based on the Dynamic Models of Robot Manipulators, and using Fourier space as motivate track come
The kinetic parameter and friction coefficient for recognizing robot, then optimize the Dynamic Models of Robot Manipulators.Wherein, base
In the theoretical joint moment that the Dynamic Models of Robot Manipulators calculates are as follows:
In formula, q,Respectively indicate robot joint angles, angular speed and angular acceleration；Indicate that robot is dynamic
The moment of inertia of mechanical model prediction；Indicate the centrifugal force and section's formula torque of Dynamic Models of Robot Manipulators prediction；
Indicate the gravitational moment that Dynamic Models of Robot Manipulators calculates；Indicate Dynamic Models of Robot Manipulators prediction
Viscous friction torque and Coulomb friction torque；D is viscous friction coefficient；μ is Coulomb friction coefficient.
When the preprocessing module recognizes the kinetic parameter and friction coefficient of robot using particle swarm algorithm, machine
The kinetic parameter of the connecting rod i of people can indicate are as follows:
λi=[mi,si,x,si,y,si,z,Ii,xx,Ii,yy,Ii,zz,Ii,xy,Ii,xz,Ii,yz,di,μi]T
In formula, i=1,2 ..., n, n are amount of articulation；miIndicate the quality of connecting rod i；si,xIndicate the mass center of connecting rod i in xiSide
To component；si,yIndicate the mass center of connecting rod i in yiThe component in direction；si,zIndicate the mass center of connecting rod i in ziThe component in direction；
Ii,xx=∫ ∫ ∫V(yi 2+zi 2)ρdυ、Ii,yy=∫ ∫ ∫V(xi 2+zi 2)ρdυ、Ii,zz=∫ ∫ ∫V(xi 2+yi 2) ρ d υ respectively indicates robot
Load is around ending coordinates axis xi,yi,ziMass mement of inertia；Around xiyi,xizi,yiziThe moment of inertia in direction is expressed as Ii,xy
=∫ ∫ ∫Vxiyiρdυ、Ii,xz=∫ ∫ ∫Vzixiρdυ、Ii,yz=∫ ∫ ∫Vyiziρ d υ, ρ indicate connecting rod density；υ indicates connecting rod volume；di
For the viscous friction coefficient of joint i；μiFor the Coulomb friction coefficient of joint i.
The impedance controller realized based on the Dynamic Models of Robot Manipulators robot end power and position it is soft
Sequence system, and location error, velocity error and the external force value that itself is calculated are transferred to the adaptive strategy control mould
Block.
The impedance controller includes position and rate control module, prediction torque feed-forward module and contact torque-feedback mould
Block.The location error and velocity error of the position and rate control module calculating robot in cartesian space, and then foundation
The location error and velocity error being calculated calculate the corresponding location error of elimination and velocity error is needed to robot end
The active force of application, then the active force is converted to the opplied moment of joint space, and then control robot motion to reduce
The location error and velocity error of robot end.
The prediction theoretical joint moment of the torque feed-forward module based on dynamics calculation machine people, and will be calculated
Theoretical joint moment is transferred to the contact torque feedback module.The contact torque feedback module is according to the motor of robot
The practical joint moment of robot is calculated in electric current, and by the practical joint moment and the theoretical joint moment come based on
The external force that robot end is subject in cartesian space is calculated, the external force is then transferred to the adaptive strategy and controls mould
Block, while controlling the robot according to the external force, to guarantee that robot outside torque when moment of face is excessive reduces
Direction movement, realizes the protection of robot.
The impedance controller realizes the position of joint space by the module and carriage transformation matrix and Jacobian matrix of robot
It sets, the conversion of speed and torque to the position of cartesian space, speed and torque, to realize to robot end position, speed
The control of degree and torque.
τ ' is the practical joint moment in robot operational process, and the impedance controller uses and directly reads servo motor
Electric current, the practical joint moment is then obtained in such a way that electric current is multiplied by torque current proportionality coefficient.Wherein, robot
Kinetics equation indicates are as follows:
In formula, τ indicates the joint driven torque calculated according to control rate；Indicate the corresponding power of prediction torque feed-forward module
Square, i.e., theoretical joint moment；Indicate position and the corresponding torque of rate control module；
xd、Respectively indicate the theoretical position and theoretical velocity of robot end；x,Respectively indicate robot end physical location and
Actual speed；Kp、KdRespectively indicate the stiffness coefficient matrix and damped coefficient matrix of impedance controller；J indicates robot Jacobi
Matrix；Indicate the corresponding torque of contact torque feedback module, τ ' indicates the practical joint calculated according to current of electric
Torque；KeIndicate contact force coefficient matrix.
Specifically, the impedance controller is first with the module and carriage transformation matrix and Jacobian matrix of robot by robot
Robot end is converted in the pose and speed of cartesian space in the angle and angular speed of joint space, and calculating robot
End generalized force；Then recycle the Jacobian matrix of robot that the theoretical joint moment of robot is calculated, it so can be with
It realizes to the pose of robot end and the control of speed.
Meanwhile the impedance controller will be calculated using the Jacobian matrix of robot based on Dynamic Models of Robot Manipulators
The practical joint moment of theoretical joint moment and robot is converted to external force suffered by robot end and torque, to obtain machine
Moment of face suffered by people end, to realize the coordinated control of robot end position and torque.
Wherein, when the theoretical joint moment calculated based on Dynamic Models of Robot Manipulators is accurate, existThen machine
People's kinetics equation can indicate are as follows:
In formula,It is related with robot end's deformation to contact torque, it can indicate are as follows:
τe=JT[γ(x-xe)]。
In formula, xeIndicate position when elastic deformation does not occur for robot end surface；γ is elastic deformation function, works as machine
When elastic deformation occurs for device people end, robot end is related with deformation quantity size by external force, therefore elastic deformation function representation
For
And then new robot dynamics' equation can be obtained:
Work as xd- x=0,x-xeWhen=0, robot actual angle, angular speed are fast with point of theory, angle respectively
Equal, deformation that end is nonelastic is spent, robot only exists prediction torque feedforward action at this time, is in equilibrium state.
The adaptive strategy control module judged according to the data received and the numerical value itself prestored, and according to
Judging result calculates controller parameter accordingly, while the controller parameter is transferred to the impedance controller, with more
The parameter of the new impedance controller, realizes the control to the impedance controller.In present embodiment, the adaptive plan
Slightly control module starts to act on when robot end is by external force, the control based on BP neural network algorithm calculating robot
Device parameter, and the obtained controller parameter is transferred to the impedance controller.
Wherein, the stationary value of robot end's location error, velocity error, power and torque error and moment of face is (preparatory
Setting value) input signal as the BP neural networkF'=(JT)-1τ is machine
The actual forces of device people end；For the theoretical power of robot end；FεFor the stationary value of robot end's external force,
According to the difference of robot end's contact environment or working forms, the stationary value of moment of face can be independently set.
Furthermore wljIndicate first of input layer of the BP neural network weight for being input to j-th of output, the output of input layer
That is the input of hidden layer is expressed as the sum of the weighting of all inputs of input layer:
The neuron of the hidden layer of BP neural network excites σ using nonlinear excitation function sigmodjIt is defeated to obtain hidden layer
σ ' outj, have:
The output of hidden layer is the input of output layer, wjkIndicate j-th of the output layer weight for being input to k-th of output, it is defeated
Layer neuron exports y outn(t) it indicates are as follows:
BP neural network output and the error e (t) and error performance function E (t) of ideal output can indicate are as follows:
E (t)=y (t)-yn(t)
The connection weight w of output layer and hidden layerjkThe formula of learning algorithm are as follows:
wjk(t+1)=wjk(t)+wjk+α(wjk(t)-wjk(t))。
Hidden layer and input layer connection weight wljThe formula of learning algorithm are as follows:
wlj(k+1)=wlj(k)+wlj+α(wlj(k)-wlj(k-1))
In formula, η is learning rate, α factor of momentum, η ∈ [0,1], α ∈ [0,1].
In present embodiment, the adaptive strategy control module in robot end there are starting to act on when contact force,
Its according to the stationary value of given moment of face come the controller parameter of periodically calculating robot, and the control that will be calculated
Device parameter is transferred to the impedance controller, to update the parameter of the impedance controller.
With specific embodiment, the present invention is further described in detail below.
Embodiment
The present embodiment is further detailed the present invention by taking 605 humanoid robot of Central China numerical control as an example, wherein Central China
The joint number n=6 of 605 humanoid robot of numerical control.Specifically, the preprocessing module is considering joint Coulomb friction power and viscous resistance
The kinetic model of 605 humanoid robot of Central China numerical control is established in the case where power, based on Newton―Leibniz formula, is based on the dynamics
Model calculates the theoretical joint moment obtained are as follows:In formula, q,Respectively
Indicate robot joint angles, angular speed and angular acceleration；Indicate the moment of inertia of kinetic model prediction；Indicate the centrifugal force and section's formula torque of kinetic model prediction；Indicate the gravitational moment that kinetic model calculates,Indicate the viscous friction power and Coulomb friction power of kinetic model prediction, d is viscous friction system
Number；μ is Coulomb friction coefficient.
In turn, the preprocessing module is based on the kinetic model, using Fourier space as excitation track, is based on grain
Swarm optimization recognizes the kinetic parameter and friction coefficient of robot, and the kinetic parameter of connecting rod i can indicate are as follows: λi=
[mi,si,x,si,y,si,z,Ii,xx,Ii,yy,Ii,zz,Ii,xy,Ii,xz,Ii,yz,di,μi]T, in formula, i=1,2 ..., 6；miIndicate connecting rod i
Quality；si,xIndicate the mass center of connecting rod i in xiThe component in direction；si,yIndicate the mass center of connecting rod i in yiThe component in direction；si,z
Indicate the mass center of connecting rod i in ziThe component in direction；Ii,xx=∫ ∫ ∫V(yi 2+zi 2)ρdυ、Ii,yy=∫ ∫ ∫V(xi 2+zi 2)ρdυ、Ii,zz
=∫ ∫ ∫V(xi 2+yi 2) ρ d υ respectively indicate robot load around ending coordinates axis xi,yi,ziMass mement of inertia；Around xiyi,xizi,
yiziThe moment of inertia in direction is expressed as Ii,xy=∫ ∫ ∫Vxiyiρdυ、Ii,xz=∫ ∫ ∫Vzixiρdυ、Ii,yz=∫ ∫ ∫Vyiziρ d υ,
Middle ρ indicates connecting rod density；υ indicates connecting rod volume；diFor the viscous friction coefficient of joint i；μiFor the Coulomb friction coefficient of joint i.
The impedance controller directly reads the electric current of servo motor, then by electric current multiplied by torque current proportionality coefficient
Mode obtain practical joint moment τ.Wherein, the kinetics equation of robot indicates are as follows:In formula, xd、Respectively indicate the theoretical position of robot end
It sets and theoretical velocity；x,Respectively indicate the physical location and actual speed of robot end；Kp、KdRespectively indicate stiffness coefficient
Matrix and damped coefficient matrix；J indicates robot Jacobian matrix；KeIndicate contact force coefficient matrix, in this example initially
Value takes:
Due to Kp、Kd、KeFor diagonal matrix, the controller parameter of impedance controller can be expressed as three matrix diagonals
18 × 1 matrixes of line element composition.
The impedance controller is using the module and carriage transformation matrix and Jacobian matrix of robot by robot in joint space
Angle and angular speed be converted to robot end in the pose and speed of cartesian space, and then robot end is calculated
Generalized force；Then Jacobian matrix is recycled to obtain the theoretical joint moment of robot, to realize the position to robot end
The control of appearance and speed.
The impedance controller closes the theoretical joint moment and the reality also according to the Jacobian matrix of robot
Section torque is converted to power suffered by robot end and torque, to obtain moment of face suffered by robot end, to realize
The coordinated control of robot end position and torque.
When the theoretical joint moment being calculated is accurate, existThen robot dynamics' equation can indicate are as follows:Wherein,To contact torque, with robot end's shape
Become related, can indicate are as follows:
τe=JT[γ(x-xe)]
In formula, xeIndicate position when elastic deformation does not occur for robot end surface；γ is elastic deformation function, works as machine
When elastic deformation occurs for device people end, robot end is related with deformation quantity size by external force, therefore elastic deformation function representation
For
And then new robot dynamics' equation can be obtained:
Work as xd- x=0,x-xeWhen=0, actual angle, angular speed and the point of theory of robot, angular speed
Equal, deformation that end is nonelastic, robot only exists prediction torque feedforward action at this time, is in equilibrium state.
In present embodiment, the adaptive strategy control module is trained using BP neural network adaptive control algorithm
The control parameter model of the impedance controller, and according to robot each moment different location error, velocity error, torque
Obtained controller parameter is transferred to the impedance control to calculate corresponding controller parameter by error and moment of face stationary value
Device processed.
Wherein, the stationary value of robot end's location error, velocity error, power and torque error and moment of face is as BP
The input signal of neural networkWherein F'=(JT)-1τ is the reality of robot end
Border power；For the theoretical power of robot end；Fε=0.4Nm is the stationary value of robot end's moment of face.This
In embodiment, wljIndicate first of input layer of the BP neural network weight for being input to j-th of output, l=24, j=18, input
The output of layer is the input of hidden layer, is expressed as the sum of the weighting of all inputs of input layer:
Wherein, the neuron of the hidden layer of BP neural network excites σ using nonlinear excitation function sigmod functionjIt obtains
Hidden layer exports σ 'j, have:
The output of hidden layer is the input of output layer, wjkIndicate j-th of the output layer weight for being input to k-th of output, this
K=18 in example, output layer neuron export yn(t) it indicates are as follows:
Network output and the error e (t) and error performance function E (t) of ideal output can indicate are as follows:
E (t)=y (t)-yn(t)
The connection weight w of output layer and hidden layerjkLearning algorithm formula are as follows:
wjk(t+1)=wjk(t)+wjk+α(wjk(t)-wjk(t))。
Hidden layer and input layer connection weight wljLearning algorithm formula are as follows:
wlj(k+1)=wlj(k)+wlj+α(wlj(k)-wlj(k-1))
In formula, η=0.8 is learning rate, and α=0.6 is factor of momentum.Robot moves Shi Yuhuan by given excitation track
Border is contacted there are contact force effect, and the adaptive strategy control module starts to act on and be calculated controller parameter, and will
Obtained controller parameter is sent to the impedance controller.
Robot adaptive impedance control system provided by the invention based on kinetic model, the control system utilize
Dynamic Models of Robot Manipulators predicts the joint of robot power and torque at each moment, prevent operation in joint moment transfinite caused by machine
Device people damage, while the time of robot accelerator can be reduced, to improve movenent performance.In addition, being based on robot power
The controller for learning model can optimize the motion profile of robot, realize high-precision control；Meanwhile the control system can be with
For being needed in the industrial robot course of work and the case where environmental interaction, by the position and torque of coordinating robot end come
Realize Shared control.
As it will be easily appreciated by one skilled in the art that the foregoing is merely illustrative of the preferred embodiments of the present invention, not to
The limitation present invention, any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should all include
Within protection scope of the present invention.
Claims (10)
1. a kind of robot adaptive impedance control system based on kinetic model, it is characterised in that:
The control system includes preprocessing module, impedance controller and adaptive strategy control module, in which:
The preprocessing module according to the Coulomb friction power and viscous force of joint of robot, using Newton―Leibniz formula for constructing
Dynamic Models of Robot Manipulators, and based on the Dynamic Models of Robot Manipulators using the dynamics ginseng of particle swarm algorithm identification robot
Several and friction coefficient, and then the Dynamic Models of Robot Manipulators is optimized；Meanwhile the preprocessing module be also used to by
Dynamic Models of Robot Manipulators after optimization is transferred to the impedance controller；
The impedance controller be used for realized according to the Dynamic Models of Robot Manipulators robot end power and position it is soft
Sequence system, and location error, velocity error and the external force value that itself is calculated are transferred to the adaptive strategy control mould
Block；
The adaptive strategy control module is used for when robot end is by external force according to the data received and itself is pre-
The numerical value deposited is judged, and correspondingly calculates controller parameter according to judging result, while the controller parameter being passed
It is defeated by the impedance controller, to update the parameter of the impedance controller, realizes the balance control of robot location and torque
System.
2. the robot adaptive impedance control system based on kinetic model as described in claim 1, it is characterised in that: institute
Stating preprocessing module is to recognize robot as excitation track using Fourier space based on the Dynamic Models of Robot Manipulators
Kinetic parameter and friction coefficient.
3. the robot adaptive impedance control system based on kinetic model as described in claim 1, it is characterised in that: base
In the theoretical joint moment that the Dynamic Models of Robot Manipulators is calculated are as follows:
In formula, q,Respectively indicate robot joint angles, angular speed and angular acceleration；Indicate robot dynamics
The moment of inertia of model prediction；Indicate the centrifugal force and section's formula torque of Dynamic Models of Robot Manipulators prediction；It indicates
The gravitational moment that Dynamic Models of Robot Manipulators calculates；Indicate the viscous of Dynamic Models of Robot Manipulators prediction
Stagnant moment of friction and Coulomb friction torque, d are viscous friction coefficient, and μ is Coulomb friction coefficient.
4. the robot adaptive impedance control system based on kinetic model as claimed in claim 3, it is characterised in that: machine
The kinetic parameter of the connecting rod i of device people are as follows:
λi=[mi,si,x,si,y,si,z,Ii,xx,Ii,yy,Ii,zz,Ii,xy,Ii,xz,Ii,yz,di,μi]T
In formula, i=1,2 ..., n, n are amount of articulation；miIndicate the quality of connecting rod i；si,xIndicate the mass center of connecting rod i in xiDirection
Component；si,yIndicate the mass center of connecting rod i in yiThe component in direction；si,zIndicate the mass center of connecting rod i in ziThe component in direction, Ii,xx=∫
∫∫V(yi 2+zi 2)ρdυ、Ii,yy=∫ ∫ ∫V(xi 2+zi 2)ρdυ、Ii,zz=∫ ∫ ∫V(xi 2+yi 2) ρ d υ respectively indicate robot load around end
Sit up straight parameter xi,yi,ziMass mement of inertia；Around xiyi,xizi,yiziThe moment of inertia in direction is expressed as Ii,xy=∫ ∫ ∫Vxiyi
ρdυ、Ii,xz=∫ ∫ ∫Vzixiρdυ、Ii,yz=∫ ∫ ∫Vyiziρdυ；ρ indicates connecting rod density；υ indicates connecting rod volume；diFor joint i's
Viscous friction coefficient；μiFor the Coulomb friction coefficient of joint i.
5. the robot adaptive impedance control system based on kinetic model as claimed in claim 3, it is characterised in that: institute
Stating impedance controller includes position and rate control module, prediction torque feed-forward module and contact torque feedback module, wherein institute
Rheme is set and rate control module is used for calculating robot in the location error and velocity error of cartesian space, and then according to meter
Obtained location error and velocity error calculates the corresponding location error of elimination and velocity error needs to apply to robot end
The active force added, then the active force is converted to the opplied moment of joint space, and then control robot motion to reduce machine
The location error and velocity error of device people end.
6. the robot adaptive impedance control system based on kinetic model as claimed in claim 5, it is characterised in that: work
When making, the prediction theoretical joint moment of the torque feed-forward module based on dynamics calculation machine people, and the reason that will be calculated
The contact torque feedback module is transferred to by joint moment；The contact torque feedback module is used for the motor according to robot
Electric current be calculated the practical joint moment of robot, and based on the practical joint moment and the theoretical joint moment come
The external force is then transferred to the adaptive strategy and controls mould by the external force that calculating robot end is subject in cartesian space
Block, while the robot is controlled according to the external force.
7. the robot adaptive impedance control system based on kinetic model as claimed in claim 6, it is characterised in that: institute
State impedance controller and realize by the module and carriage transformation matrix and Jacobian matrix of robot position, speed and the power of joint space
Square to the position of cartesian space, speed and torque conversion, to realize control to robot end position, speed and torque
System.
8. special such as the described in any item robot adaptive impedance control systems based on kinetic model of claim 1-7
Sign is: the kinetics equation of robot are as follows:
In formula, J indicates robot Jacobian matrix；Kp、KdRespectively indicate the stiffness coefficient matrix and damped coefficient of impedance controller
Matrix；xd、Respectively indicate the theoretical position and theoretical velocity of robot end；x,Respectively indicate the reality of robot end
Position and actual speed；KeIndicate contact force coefficient matrix；xeIndicate position when elastic deformation does not occur for robot end surface
It sets.
9. special such as the described in any item robot adaptive impedance control systems based on kinetic model of claim 1-7
Sign is: the adaptive strategy control module starts to act on when robot end has external force, is calculated based on BP neural network
The controller parameter of method calculating robot, and the obtained controller parameter is transferred to the impedance controller.
10. the robot adaptive impedance control system based on kinetic model as claimed in claim 9, it is characterised in that:
The stationary value of robot end's location error, velocity error, power and torque error and moment of face is as the BP neural network
Input signalF'=(JT)-1τ is the actual forces of robot end；For the theoretical power of robot end, FεFor the stationary value of robot end's external force；xd、Respectively indicate machine
The theoretical position and theoretical velocity of people end；x,Respectively indicate the physical location and actual speed of robot end.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910352004.9A CN110065070B (en) | 2019-04-29 | 2019-04-29 | Robot self-adaptive impedance control system based on dynamic model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910352004.9A CN110065070B (en) | 2019-04-29 | 2019-04-29 | Robot self-adaptive impedance control system based on dynamic model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110065070A true CN110065070A (en) | 2019-07-30 |
| CN110065070B CN110065070B (en) | 2020-09-18 |
Family
ID=67369374
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910352004.9A Active CN110065070B (en) | 2019-04-29 | 2019-04-29 | Robot self-adaptive impedance control system based on dynamic model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110065070B (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110561438A (en) * | 2019-09-19 | 2019-12-13 | 华中科技大学 | Industrial robot manpower/position compliance control method based on kinetic parameter identification |
| CN111037571A (en) * | 2019-12-31 | 2020-04-21 | 广东工业大学 | Robot self-adaptive variable damping impedance control method |
| CN111152213A (en) * | 2019-12-05 | 2020-05-15 | 北京蒂斯科技有限公司 | Mechanical arm vibration compensation method and device based on hybrid control |
| CN111230867A (en) * | 2020-01-16 | 2020-06-05 | 腾讯科技（深圳）有限公司 | Robot motion control method, motion control equipment and robot |
| CN111283687A (en) * | 2020-03-17 | 2020-06-16 | 北京理工大学 | Robot joint position control system and feedback compensation method of dynamic moment thereof |
| CN111531543A (en) * | 2020-05-12 | 2020-08-14 | 中国科学院自动化研究所 | Robot self-adaptive impedance control method based on biological heuristic neural network |
| CN111590165A (en) * | 2020-06-04 | 2020-08-28 | 北部湾大学 | Ship assemblage plate welding robot based on remote correction and welding method |
| CN111673733A (en) * | 2020-03-26 | 2020-09-18 | 华南理工大学 | Intelligent self-adaptive compliance control method of robot in unknown environment |
| CN111890370A (en) * | 2020-08-15 | 2020-11-06 | 前元运立(北京)机器人智能科技有限公司 | Robot power control method and device applying double encoders |
| CN112720460A (en) * | 2020-12-07 | 2021-04-30 | 深圳市优必选科技股份有限公司 | Robot control method, device, computer readable storage medium and robot |
| CN112743538A (en) * | 2019-10-30 | 2021-05-04 | 皮尔茨公司 | Model predictive interaction control |
| CN112809666A (en) * | 2020-12-17 | 2021-05-18 | 安徽工业大学 | 5-DOF mechanical arm force and position tracking algorithm based on neural network |
| CN113128018A (en) * | 2019-12-31 | 2021-07-16 | 深圳市优必选科技股份有限公司 | Friction force calculation method and device, robot and readable storage medium |
| CN113319857A (en) * | 2021-08-03 | 2021-08-31 | 季华实验室 | Mechanical arm force and position hybrid control method and device, electronic equipment and storage medium |
| CN113500597A (en) * | 2021-07-08 | 2021-10-15 | 南京工业大学 | Multi-end teleoperation sliding mode impedance control method based on force translation mechanism |
| CN114012734A (en) * | 2021-12-03 | 2022-02-08 | 西安交通大学 | Parameter-adaptive robot collision detection method |
| CN114516052A (en) * | 2022-03-23 | 2022-05-20 | 杭州湖西云百生科技有限公司 | Dynamics control method and system of parallel real-time high-performance multi-axis mechanical arm |
| CN114619437A (en) * | 2020-12-08 | 2022-06-14 | 山东新松工业软件研究院股份有限公司 | Flexible impedance control method for redundant multi-joint robot |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104626152A (en) * | 2013-11-14 | 2015-05-20 | 沈阳新松机器人自动化股份有限公司 | Active compliant control method and device of industrial robot |
| US20150202775A1 (en) * | 2012-01-17 | 2015-07-23 | Seiko Epson Corporation | Robot controller, robot system, robot control method |
| KR20150129225A (en) * | 2014-05-09 | 2015-11-19 | 현대자동차주식회사 | Method and system for sensing motion intent of wearable robot |
| CN105479459A (en) * | 2015-12-29 | 2016-04-13 | 深圳市汇川技术股份有限公司 | Zero-force control method and system for robot |
| CN107139171A (en) * | 2017-05-09 | 2017-09-08 | 浙江工业大学 | A kind of industrial robot collision free trajectory method based on Torque Control |
-
2019
- 2019-04-29 CN CN201910352004.9A patent/CN110065070B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150202775A1 (en) * | 2012-01-17 | 2015-07-23 | Seiko Epson Corporation | Robot controller, robot system, robot control method |
| CN104626152A (en) * | 2013-11-14 | 2015-05-20 | 沈阳新松机器人自动化股份有限公司 | Active compliant control method and device of industrial robot |
| KR20150129225A (en) * | 2014-05-09 | 2015-11-19 | 현대자동차주식회사 | Method and system for sensing motion intent of wearable robot |
| CN105479459A (en) * | 2015-12-29 | 2016-04-13 | 深圳市汇川技术股份有限公司 | Zero-force control method and system for robot |
| CN107139171A (en) * | 2017-05-09 | 2017-09-08 | 浙江工业大学 | A kind of industrial robot collision free trajectory method based on Torque Control |
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110561438B (en) * | 2019-09-19 | 2021-02-05 | 华中科技大学 | Industrial robot manpower/position compliance control method based on kinetic parameter identification |
| CN110561438A (en) * | 2019-09-19 | 2019-12-13 | 华中科技大学 | Industrial robot manpower/position compliance control method based on kinetic parameter identification |
| CN112743538A (en) * | 2019-10-30 | 2021-05-04 | 皮尔茨公司 | Model predictive interaction control |
| CN111152213A (en) * | 2019-12-05 | 2020-05-15 | 北京蒂斯科技有限公司 | Mechanical arm vibration compensation method and device based on hybrid control |
| CN111152213B (en) * | 2019-12-05 | 2021-03-16 | 北京蒂斯科技有限公司 | Mechanical arm vibration compensation method and device based on hybrid control |
| CN111037571B (en) * | 2019-12-31 | 2022-12-16 | 广东工业大学 | Robot self-adaptive variable damping impedance control method |
| CN111037571A (en) * | 2019-12-31 | 2020-04-21 | 广东工业大学 | Robot self-adaptive variable damping impedance control method |
| CN113128018A (en) * | 2019-12-31 | 2021-07-16 | 深圳市优必选科技股份有限公司 | Friction force calculation method and device, robot and readable storage medium |
| CN111230867A (en) * | 2020-01-16 | 2020-06-05 | 腾讯科技（深圳）有限公司 | Robot motion control method, motion control equipment and robot |
| CN111283687A (en) * | 2020-03-17 | 2020-06-16 | 北京理工大学 | Robot joint position control system and feedback compensation method of dynamic moment thereof |
| CN111673733A (en) * | 2020-03-26 | 2020-09-18 | 华南理工大学 | Intelligent self-adaptive compliance control method of robot in unknown environment |
| CN111673733B (en) * | 2020-03-26 | 2022-03-29 | 华南理工大学 | Intelligent self-adaptive compliance control method of robot in unknown environment |
| CN111531543A (en) * | 2020-05-12 | 2020-08-14 | 中国科学院自动化研究所 | Robot self-adaptive impedance control method based on biological heuristic neural network |
| CN111531543B (en) * | 2020-05-12 | 2021-10-08 | 中国科学院自动化研究所 | Robot self-adaptive impedance control method based on biological heuristic neural network |
| CN111590165A (en) * | 2020-06-04 | 2020-08-28 | 北部湾大学 | Ship assemblage plate welding robot based on remote correction and welding method |
| CN111890370A (en) * | 2020-08-15 | 2020-11-06 | 前元运立(北京)机器人智能科技有限公司 | Robot power control method and device applying double encoders |
| CN112720460A (en) * | 2020-12-07 | 2021-04-30 | 深圳市优必选科技股份有限公司 | Robot control method, device, computer readable storage medium and robot |
| CN112720460B (en) * | 2020-12-07 | 2022-06-10 | 深圳市优必选科技股份有限公司 | Robot control method, device, computer readable storage medium and robot |
| CN114619437A (en) * | 2020-12-08 | 2022-06-14 | 山东新松工业软件研究院股份有限公司 | Flexible impedance control method for redundant multi-joint robot |
| CN112809666A (en) * | 2020-12-17 | 2021-05-18 | 安徽工业大学 | 5-DOF mechanical arm force and position tracking algorithm based on neural network |
| CN113500597A (en) * | 2021-07-08 | 2021-10-15 | 南京工业大学 | Multi-end teleoperation sliding mode impedance control method based on force translation mechanism |
| CN113319857A (en) * | 2021-08-03 | 2021-08-31 | 季华实验室 | Mechanical arm force and position hybrid control method and device, electronic equipment and storage medium |
| CN113319857B (en) * | 2021-08-03 | 2021-10-12 | 季华实验室 | Mechanical arm force and position hybrid control method and device, electronic equipment and storage medium |
| CN114012734A (en) * | 2021-12-03 | 2022-02-08 | 西安交通大学 | Parameter-adaptive robot collision detection method |
| CN114516052A (en) * | 2022-03-23 | 2022-05-20 | 杭州湖西云百生科技有限公司 | Dynamics control method and system of parallel real-time high-performance multi-axis mechanical arm |
| CN114516052B (en) * | 2022-03-23 | 2023-12-22 | 杭州湖西云百生科技有限公司 | Dynamics control method and system for parallel real-time high-performance multi-axis mechanical arm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110065070B (en) | 2020-09-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110065070A (en) | A kind of robot adaptive impedance control system based on kinetic model | |
| He et al. | Design and adaptive control for an upper limb robotic exoskeleton in presence of input saturation | |
| CN109848983B (en) | Method for guiding robot to cooperatively work by high-compliance person | |
| Xu et al. | Trajectory tracking control of omnidirectional wheeled mobile manipulators: robust neural network-based sliding mode approach | |
| Pradhan et al. | Real-time adaptive control of a flexible manipulator using reinforcement learning | |
| Li et al. | An input dead zones considered adaptive fuzzy control approach for double pendulum cranes with variable rope lengths | |
| CN110936382B (en) | Data-driven industrial robot energy consumption optimization method | |
| CN108789417A (en) | Reconfigurable manipulator based on adaptive Dynamic Programming protects cost decentralized control method | |
| CN109839826A (en) | Double drive coordinated movement of various economic factors synchronisation control means | |
| CN109514564A (en) | A kind of compound quadratic form multi-joint mechanical arm method for optimally controlling | |
| Jamisola et al. | Task-space modular dynamics for dual-arms expressed through a relative jacobian | |
| Bruckmann et al. | An energy-efficient wire-based storage and retrieval system | |
| CN110597051A (en) | Stewart stable platform control method based on RBF neural network | |
| Sakaino et al. | Imitation learning for variable speed contact motion for operation up to control bandwidth | |
| CN108693776A (en) | A kind of robust control method of Three Degree Of Freedom Delta parallel robots | |
| CN117452966A (en) | Unmanned aerial vehicle force collaborative hanging and carrying method based on single-axis tension sensor | |
| Tong et al. | Neural network based visual servo control under the condition of heavy loading | |
| Hu et al. | Motion and force control of coordinated robots during constrained motion tasks | |
| Nagata et al. | Adaptive learning with large variability of teaching signals for neural networks and its application to motion control of an industrial robot | |
| Ma et al. | Unknown constrained mechanisms operation based on dynamic hybrid compliance control | |
| Ramos et al. | Nonlinear regulation of an underactuated system | |
| CN110488608B (en) | Intelligent kinetic parameter identification method and module for driving and controlling integrated control system | |
| Murao et al. | Predictive visual feedback control with eye-in-hand system via stabilizing receding horizon approach | |
| Yang et al. | Dynamic compensation control of flexible macro–micro manipulator systems | |
| CN114750148A (en) | Force closed loop zero force control method and system for self-adaptive measurement of gravity |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |