CN115609585A - Control method and system for tail end position of continuous body robot arm and robot arm system - Google Patents

Control method and system for tail end position of continuous body robot arm and robot arm system Download PDF

Info

Publication number
CN115609585A
CN115609585A CN202211280925.7A CN202211280925A CN115609585A CN 115609585 A CN115609585 A CN 115609585A CN 202211280925 A CN202211280925 A CN 202211280925A CN 115609585 A CN115609585 A CN 115609585A
Authority
CN
China
Prior art keywords
robot arm
neural network
control
driving
continuum robot
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.)
Pending
Application number
CN202211280925.7A
Other languages
Chinese (zh)
Inventor
齐鹏
刘文杰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tongji University
Original Assignee
Tongji University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tongji University filed Critical Tongji University
Priority to CN202211280925.7A priority Critical patent/CN115609585A/en
Publication of CN115609585A publication Critical patent/CN115609585A/en
Pending legal-status Critical Current

Links

Images

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
    • 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
    • B25J9/1607—Calculation of inertia, jacobian matrixes and inverses
    • 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
    • B25J9/161—Hardware, e.g. neural networks, fuzzy logic, interfaces, processor
    • 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
    • B25J9/1656—Programme controls characterised by programming, planning systems for manipulators
    • B25J9/1664—Programme controls characterised by programming, planning systems for manipulators characterised by motion, path, trajectory planning

Landscapes

  • Engineering & Computer Science (AREA)
  • Robotics (AREA)
  • Mechanical Engineering (AREA)
  • Automation & Control Theory (AREA)
  • Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Fuzzy Systems (AREA)
  • Software Systems (AREA)
  • Feedback Control In General (AREA)
  • Manipulator (AREA)

Abstract

The invention discloses a control method and a control system for the tail end position of a continuum robot arm and a robot arm system, wherein the control method comprises the steps of training a neural network model of the inverse kinematics of the continuum robot arm; obtaining a driving variable quantity required by control through the pre-trained neural network, and controlling the tail end position of the continuum robot arm to move to the expected position according to the driving variable quantity output by the neural network; measuring the actual position of the tail end of the continuum robot arm in real time through a position sensor, and calculating a tail end position control error; and if the control error of the tail end position is larger than a set threshold value, predicting the real-time Jacobian matrix variation and the driving compensation amount through an online training neural network. The invention achieves the purpose of updating the Jacobian matrix by training the output value of the neural network on line, calculates the driving compensation quantity of the current continuum robot by the updated Jacobian matrix, and further realizes the high-precision control of the continuum robot.

Description

Control method and system for tail end position of continuous body robot arm and robot arm system
Technical Field
The invention relates to the technical field of robots, in particular to a method and a system for controlling the tail end position of a continuum robot arm and a robot arm system.
Background
The continuum robot is a novel robot based on bionics of biological organs such as octopus tentacles and elephants noses, does not have rigid connecting rods and discrete joints, has excellent flexibility and flexibility, has very wide application prospects in the fields of minimally invasive intervention operations, narrow and small environment detection and rescue, space operation and the like, and gradually becomes a new research hotspot. However, the characteristics of flexibility and super redundancy also bring great test to the modeling and control of the continuum robot.
Unlike a motion-predictable rigid body robot, a continuous body robot has infinite degrees of freedom, but is susceptible to deformation when subjected to external loads. The control task of a continuum robot is therefore more challenging and difficult than with a traditional robot hand. Nowadays, precise position control of a continuum robot is one of the most important research topics for the continuum robot.
At present, a continuum control method is mainly closed-loop feedback control with a model. Researchers have discovered new inverse kinematics models for continuum robots that to some extent capture the ability to bend and bend. Such inverse kinematics models are constant curvature models. The control method depending on the specific inverse kinematics model is the model-based control. The feedback control method based on the continuum robot inverse kinematics model has obvious defects. When the continuum robot works in a limited environment or is interfered by external force, the Jacobian matrix based on the inverse kinematics model is incorrect. Therefore, under unknown environments, we cannot model the configuration and jacobian matrix of a continuum robot efficiently.
Therefore, it is an urgent technical problem to develop a method and a system for precisely controlling the position of the end of the continuum robot arm.
Disclosure of Invention
Due to the defects in the prior art, the invention provides a control method and a control system for the tail end position of a continuum robot arm and a robot arm system. And calculating the current driving compensation quantity of the continuum robot through the updated Jacobian matrix, thereby realizing the high-precision control of the continuum robot.
To achieve the above object, in one aspect, the present invention provides a method for controlling a position of an end of a continuum robot arm, comprising the steps of:
s1, training a neural network model of continuous body robot arm inverse kinematics;
s2, inputting a given control expected position, and obtaining a driving variable quantity required by control through the pre-trained neural network; controlling the end position of the continuum robot arm to move to the expected position according to the driving change quantity output by the neural network;
after the steps S1 and S2 are executed, the method is characterized by further comprising the following steps:
s3, measuring the actual position of the tail end of the continuous body robot arm in real time through a position sensor, and calculating a tail end position control error;
s4, judging whether the control error of the tail end position is larger than a set threshold value or not; if so, indicating that the control precision is insufficient, and entering the next step; otherwise, the control progress meets the control requirement, and the next target position can be input for control;
and S5, predicting the real-time jacobian matrix variation through an online training neural network, further calculating a driving compensation amount through the calculated jacobian matrix and the position control error of the current moment, and driving the tail end position of the robot to change according to the driving compensation amount, so that the control precision meets the control requirement.
The method combines the pre-trained neural network, real-time terminal position measurement and on-line neural network training to accurately obtain the driving compensation quantity of the terminal position of the continuum robot arm in time. On-line neural network training does not need a training set and a testing set, input and output are measured values of a position sensor at the last moment and the current moment, and a predicted label value is a set position control target value.
Further, the driving motor rotates for corresponding circles to drive the two driving ropes in the continuous body mechanical arm to linearly displace; the tail end of the continuous mechanical arm moves along with the change of the linear displacement of the driving rope; and the driving motor executes operation according to the acquired driving variable quantity instruction.
Further, in step S3, a magnetic sensor is disposed at the front end of the continuum manipulator arm, and an actual position of the tail end of the continuum manipulator arm is measured in real time through interaction between the magnetic sensor and an external magnetic field generator.
Further, in step S5, the online training neural network is a 3-layer structure of an input layer, a hidden layer, and an output layer: the input layer is provided with 4 neurons, the hidden layer is provided with 12 neurons, and the output layer is provided with 4 neurons; and obtaining the variation of the Jacobian matrix and the driving compensation quantity by adopting the neural network of the BP algorithm.
Further, the loss function L selected by the online training neural network is:
Figure BDA0003898309160000031
where P (k) is the actual position of the end of the continuum robot at time k, P d To desired control position, [ P (k) -P d ] T The error length vector is controlled for the position at time k.
Further, calculating the variation of each parameter during sub-optimization by solving the partial derivative of the loss function L on the parameters of the online training neural network, and optimizing the parameters, so as to train the parameters circularly, update a new Jacobian matrix, obtain the required driving compensation amount, and finally enable the control error of the tail end position of the continuum robot arm to be smaller than a set value.
In another aspect, the present invention provides a system for controlling a position of an end of a continuum robot arm, comprising: the pre-training neural network module is configured to pre-train and obtain the inverse kinematics model of the continuum robot arm to obtain the required driving variation of the target position at the tail end of the continuum robot arm;
a position sensor sensing module configured to obtain an actual position of a tip of the continuum robotic arm;
the online training neural network module is configured to train in real time and obtain a driving compensation quantity between the real-time displacement of the tail end of the continuum robot arm and the inverse kinematics model;
and the driving module is configured to drive the continuous body robot arm to act according to the driving variation obtained by the pre-trained neural network module and the driving compensation obtained by the online-trained neural network module.
In another aspect, the present invention provides an electronic device, which includes a memory, a processor, and computer instructions stored in the memory and executed on the processor, wherein the computer instructions, when executed by the processor, implement the method for controlling the position of the end of the continuum robot arm.
In a final aspect, the present invention provides a robot arm system, including a base, a continuum robot arm body having one end fixedly connected to the base, a driving rope in the continuum robot arm, a driving motor for the driving rope, and a position sensor at the end of the continuum robot arm, and further including a control system for controlling the position of the end of the continuum robot arm.
Compared with the prior art, the invention has the following advantages or beneficial effects:
the invention combines the real-time measured tail end position of the robot arm and the online training neural network to obtain the driving compensation quantity of the tail end position of the continuum robot arm, and is a method for realizing closed-loop control of the continuum robot arm based on model-free online training neural network to further update a Jacobian matrix. The method solves the control problem of the continuum robot in a limited environment, further improves the control precision of the tail end position of the robot, improves the control accuracy, and provides a better control strategy for the continuum robot in the control field.
Drawings
The invention and its features, aspects and advantages will become more apparent from the following detailed description of non-limiting embodiments, which is to be read in connection with the accompanying drawings. Like reference symbols in the various drawings indicate like elements. The drawings are not necessarily to scale, emphasis instead being placed upon illustrating the principles of the invention.
FIG. 1 is a top view of a continuum robot arm in an embodiment of the invention;
FIG. 2 is a side view of an experimental arrangement according to an embodiment of the present invention;
FIG. 3 is a flow chart illustrating steps of a method for controlling the position of the end of a continuum robot in accordance with one embodiment of the invention;
FIG. 4 depicts the meaning of continuum robot arm end position control errors in an embodiment of the present invention;
FIG. 5 is a control flow diagram of a method for controlling the end position of a continuum robot arm in accordance with an embodiment of the present invention;
wherein, 1, the continuum mechanical arm body; 2. a first drive rope; 3. a second drive rope; 4. a base; 5. a magnetic sensor; 6. a magnetic field generator.
Detailed Description
The invention will be further described with reference to the following drawings and specific examples, which are not intended to limit the invention thereto.
Example 1
Referring to fig. 1, a basic structure of a continuum robot arm includes a continuum robot arm body 1 and a drive rope connected to a base 4. As a preferred embodiment, assuming that the number of the drive ropes is two, that is, the first drive rope 2 and the second drive rope 3, the drive principle is to bend the continuum arm body 1 by changing the lengths of the two drive ropes in the continuum arm body. Since there are only 2 drive ropes, the drive space here refers to the length vector of the drive ropes passing through the continuum robot, i.e., Q = [ Q ] ([ Q ]) 1 ,q 2 ] T Where Q is a drive length vector, Q 1 For the first drive rope length (cm), q 2 For a second drive rope length (cm). It follows that the continuum robotic arm moves with only 2 degrees of freedom. The task space here refers to a two-dimensional position of the end of the continuum robot arm, and the origin of the two-dimensional position coordinates is located at the center of the bottom of the continuum robot arm.
Referring to fig. 2 and 3, based on the above models, a method for controlling the position of the end of a continuum robot arm includes the steps of:
s1, training a neural network model of inverse kinematics of a continuous body robot arm;
s2, inputting a given control expected position, and obtaining a driving variable quantity required by control through the pre-trained neural network; and controlling the end position of the continuum robot arm to move to the expected position according to the driving change quantity output by the neural network.
Inputting a desired position of a given control, namely assigning a value to a set desired position variable; a pre-trained neural network was used as an inverse kinematics model for the continuum manipulator of this example. The desired position variable is converted into a driving variation amount required for control by a pre-trained neural network. The neural network training process is as follows: first, the task space (P) of each position needs to be obtained through experiments 1 ,P 2 ,P 3 ....P n ) And its corresponding driving space (Q) 1 ,Q 2 ,Q 3 ....Q n ) The task space data is used as training input, and the driving space is used as target output. And performing regression prediction on the data by using a BP algorithm until the test error is lower than a set value. After the neural network parameter training is completed, the neural network parameter training can be approximated to a mapping function Q = Φ (P), where P is a desired control position and Q is a predicted value of a driving quantity corresponding to the desired position.
In step S2 of this embodiment, first, the driving variation Δ Q output by the neural network in the previous step is obtained, and then the driving variation Δ Q is input as a control signal, and then the driving motor is correspondingly changed, and the driving motor rotates for a corresponding number of turns to drive the driving rope to linearly displace. Finally, as the drive rope length changes, the continuous robot arm end position moves to a desired position.
And S3, measuring the actual position of the tail end of the continuum robot arm in real time through the position sensor, and calculating a tail end position control error. The position sensor can select the magnetic sensor 5 arranged at the front end of the continuum mechanical arm, and the actual position of the tail end of the continuum mechanical arm is measured in real time through interaction of the magnetic sensor 5 and an external magnetic field generator 6. The magnetic field generator 6 can be a medical magnetic field generator of NDIAurera V3 series, and the magnetic sensor 5 can be a magnetic positioning coil matched with the magnetic field generator. It will be appreciated that the position sensor may also be another type or model of in vivo positioning sensor to enable positioning of the end position of the continuum manipulator arm.
Referring to FIG. 4, P d Is the desired control position, Δ P (k) refers to the control error value at time k, Δ P r (k) And measuring the actual position of the tail end of the mechanical arm of the continuum at the k moment by using a position sensor, uploading position information in real time and calculating the actual tail end position control error delta P (k), wherein k refers to the current time line.
S4, judging whether the tail end position control error delta P (k) is larger than a set threshold value L or not; if so, indicating that the control precision is insufficient and the driving quantity compensation control is needed, entering the next step; otherwise, the control progress reaches the control requirement, and the next target position can be input for control.
And S5, predicting the real-time jacobian matrix variation through an online training neural network, further calculating a driving compensation amount through the calculated jacobian matrix and the position control error of the current moment, and driving the tail end position of the robot to change according to the driving compensation amount, so that the control precision meets the control requirement.
Referring to fig. 5, in order to obtain the driving compensation amount, the method predicts the change amount of the jacobian matrix at the time by training the neural network on line, and then obtains the driving compensation amount Δ q from the calculated jacobian matrix and the position control error Δ P (k) at the current time c ，Δq c The calculation formula is as follows:
ΔP(k)＝P(k)-P d
ΔP r (k)＝P(k+1)-P(k)
J(k)＝J(k-1)+ΔJ(k)
Δq c (k)＝J(k)ΔP(k)
where J (k) is the Jacobian matrix value predicted by the neural network at time k, P d Is the desired control position, Δ P (k) refers to the control error value at time k, Δ P r (k) The k time of the finger is the actual displacement of the end of the arm.
Before predicting the variation of the Jacobian matrix through a neural network, the Jacobian matrix needs to be initialized, so that an initialized Jacobian matrix J is obtained 0 。
The jacobian matrix initialization method is as follows:
before experimental control, the continuum robot arm is started, and a tiny task space movement amount P = [ delta x, delta z ] is given] T So that a variation Q = [ Δ Q ] of the corresponding driving space is obtained 1 ,Δq 2 ] T 。
Calculating the mapping relation of the task space movement amount and the corresponding driving space variation amount, and taking the mapping matrix as an initialized Jacobian matrix J 0 As follows.
Figure BDA0003898309160000081
Q＝J 0 P
The training process of the online real-time training neural network is as follows:
the online training neural network has 3 layers, 4 neurons in the input layer, and P as input value 1 (k)-P d (1) ，P 2 (k)-P d (2) ，P 1 (k)-P 1 (k-1)，P 2 (k)-P 2 (k-1). The hidden layer is provided with 12 neurons. The output layer is provided with 4 neurons, and the output values are respectively delta J 11 ,ΔJ 12 ,ΔJ 21 ,ΔJ 22 。
Wherein
Figure BDA0003898309160000082
The activation function of the hidden layer is: g [ = [ exp (x) -exp (-x) ]/[ exp (x) + exp (-x) ]
The activation function of the output layer is: h [ · ] = [ exp (x) ]/[ exp (x) + exp (-x) ]
Selected loss function L:
Figure BDA0003898309160000083
wherein the desired control position P d Is a label.
The neural network BP algorithm can obtain:
the hidden layer output is:
Figure BDA0003898309160000084
wherein: x is the number of 1 ,x 2 ,x 3 ,x 4 ，w ji (1) And b j (1) Weights and biases for the hidden layers.
The output of the output layer is:
Figure BDA0003898309160000085
wherein: w is a rj (2) And b r (2) The weights and offsets of the output layers.
Jacobian matrix variance:
Figure BDA0003898309160000086
Figure BDA0003898309160000091
the drive compensation amount is:
Figure BDA0003898309160000092
Figure BDA0003898309160000093
wherein the content of the first and second substances,
Figure BDA0003898309160000094
the k-time drive amount is:
Figure BDA0003898309160000095
drive compensation amount Δ q c (k) Acting on the continuum mechanical arm, wherein the position of the end effector is P (k + 1), and the variation is delta P r (k) In that respect And order
Figure BDA0003898309160000096
Training boundary conditions of the online neural network: l (k) < L or the training times epochs > = n, wherein L is the error temporaryAnd n is the set training times.
Next, to reduce the error, the neural network parameters need to be optimized, i.e., w ji (1) ,b j (1) ,w ri (2) ,b r (2) ,. So that the loss function L is less than the critical value, the loss function L:
Figure BDA0003898309160000097
and calculating the partial derivatives of the parameters by the loss function L, and further calculating the parameter variation at each suboptimum time.
The parameter optimization formula of the online neural network is as follows:
Figure BDA0003898309160000098
Figure BDA0003898309160000099
Figure BDA00038983091600000910
Figure BDA00038983091600000911
Figure BDA0003898309160000101
Figure BDA0003898309160000102
Figure BDA0003898309160000103
Figure BDA0003898309160000104
Figure BDA0003898309160000105
wherein: when the ratio of r =1, the ratio,
Figure BDA0003898309160000106
when r =2, the signal is transmitted,
Figure BDA0003898309160000107
when r =3, the signal is transmitted,
Figure BDA0003898309160000108
when r =4, the signal is transmitted,
Figure BDA0003898309160000109
Figure BDA00038983091600001010
wherein
Figure BDA00038983091600001011
Figure BDA00038983091600001012
Wherein
Figure BDA00038983091600001013
Thus, the parameter optimization formula is used for training the neural network parameters circularly, a new Jacobian matrix is updated, the required driving compensation amount is obtained, and finally the tail end position of the continuum robot arm is changed until the position control error is smaller than the set value L.
Example 2
A control system for a position of an end of a continuum robot arm, comprising: the pre-trained neural network module is configured to pre-train and obtain the inverse kinematics model of the continuum robot arm to obtain the required driving variation of the target position at the tail end of the continuum robot arm;
a position sensor sensing module configured to obtain an actual position of a tip of the continuum robotic arm;
the online training neural network module is configured to train in real time and obtain a driving compensation quantity between the real-time displacement of the tail end of the continuum robot arm and the inverse kinematics model;
and the driving module is configured to drive the continuous body robot arm to act according to the driving variation obtained by the pre-trained neural network module and the driving compensation obtained by the online-trained neural network module.
The control system for the end position of the continuum robot arm may implement the control method described in embodiment 1.
Example 3
An electronic device comprising a memory and a processor and computer instructions stored on the memory and executed on the processor, wherein the computer instructions, when executed by the processor, perform the method of controlling the position of the end of a continuum robot arm of embodiment 1.
Example 4
Referring to fig. 2, a robot arm system includes a base 4, a continuum robot arm body 1 having one end fixedly connected to the base, a drive rope (only the first drive rope 2 is shown) in the continuum robot arm, a drive motor (not shown) of the drive rope, and a magnetic sensor 5 at the end of the continuum robot arm, and further includes a control system for controlling the position of the end of the continuum robot arm according to embodiment 1.
In summary, the present invention provides a method and a system for controlling the position of the end of a continuum robot arm, and a robot arm system, including training a neural network model of inverse kinematics of the continuum robot arm; obtaining a driving variable quantity required by control through the pre-trained neural network, and controlling the tail end position of the continuum robot arm to move to the expected position according to the driving variable quantity output by the neural network; measuring the actual position of the tail end of the continuum robot arm in real time through a position sensor, and calculating a tail end position control error; and if the control error of the tail end position is larger than a set threshold value, predicting the real-time Jacobian matrix variation and the driving compensation amount through an online training neural network. The invention achieves the purpose of updating the Jacobian matrix by training the output value of the neural network on line, calculates the driving compensation quantity of the current continuum robot by the updated Jacobian matrix, and further realizes the high-precision control of the continuum robot.
Those skilled in the art will appreciate that variations may be implemented by those skilled in the art in combination with the prior art and the above-described embodiments, and will not be described herein in detail. Such variations do not affect the essence of the present invention, and are not described herein.
The above description is of the preferred embodiment of the invention. It is to be understood that the invention is not limited to the particular embodiments described above, in which devices and structures not described in detail are understood to be implemented in a manner that is conventional in the art; those skilled in the art can make many possible variations and modifications to the disclosed solution, or modify the equivalent embodiments with equivalent variations, without departing from the scope of the solution, without thereby affecting the spirit of the invention. Therefore, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention are still within the scope of the protection of the technical solution of the present invention, unless the contents of the technical solution of the present invention are departed.
Those of ordinary skill in the art will appreciate that the various illustrative elements, i.e., algorithm steps, described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

Claims (9)

1. A method of controlling the position of the end of a continuum robot arm, comprising the steps of:
s1, training a neural network model of continuous body robot arm inverse kinematics;
s2, inputting a given control expected position, and obtaining a driving variable quantity required by control through the pre-trained neural network; controlling the end position of the continuum robot arm to move to the expected position according to the driving change quantity output by the neural network;
s3, measuring the actual position of the tail end of the continuous body robot arm in real time through a position sensor, and calculating a tail end position control error;
s4, judging whether the control error of the tail end position is larger than a set threshold value or not; if so, indicating that the control precision is insufficient, and entering the next step; otherwise, the control progress reaches the control requirement, and the next target position can be input for control;
and S5, predicting the real-time jacobian matrix variation through an online training neural network, further calculating a driving compensation amount through the calculated jacobian matrix and the position control error at the current moment, and driving the tail end position of the robot to change according to the driving compensation amount so as to enable the control precision to meet the control requirement.
2. The method for controlling the position of the tail end of the continuum robot arm according to claim 1, wherein the driving motor rotates for a corresponding number of turns to drive the two driving ropes in the continuum robot arm to linearly displace; the tail end of the continuous mechanical arm moves along with the change of the linear displacement of the driving rope; and the driving motor executes operation according to the acquired driving variable quantity instruction.
3. The method according to claim 1, wherein in step S3, the front end of the continuum robot arm is provided with a magnetic sensor, and the actual position of the end of the continuum robot arm is measured in real time through interaction between the magnetic sensor and an external magnetic field generator.
4. The method for controlling the end position of the continuum robot arm according to claim 1, wherein in the step S5, the online training neural network is a 3-layer structure of an input layer, a hidden layer and an output layer: the input layer is provided with 4 neurons, the hidden layer is provided with 12 neurons, and the output layer is provided with 4 neurons; and obtaining the variation of the Jacobian matrix and the driving compensation quantity by adopting the neural network of the BP algorithm.
5. The method as claimed in claim 4, wherein the loss function L selected by the on-line training neural network is:
Figure FDA0003898309150000021
where P (k) is the actual position of the end of the continuum robot at time k, P d To desired control position, [ P (k) -P d ] T The error length vector is controlled for the position at time k.
6. The method as claimed in claim 5, wherein the variation of the parameters at each sub-optimization is calculated by solving the partial derivatives of the loss function L to the parameters of the on-line training neural network, and the parameters are optimized, so as to train the parameters cyclically, update a new Jacobian matrix, obtain the required driving compensation amount, and finally make the control error of the end position of the continuum robot arm smaller than the set value.
7. A control system for a position of an end of a continuum robot arm, comprising:
the pre-training neural network module is configured to pre-train and obtain the inverse kinematics model of the continuum robot arm to obtain the required driving variation of the target position at the tail end of the continuum robot arm;
a position sensor sensing module configured to obtain an actual position of a tip of the continuum robotic arm;
the online training neural network module is configured to train in real time and obtain a driving compensation quantity between the real-time displacement of the tail end of the continuum robot arm and the inverse kinematics model;
and the driving module is configured to drive the continuous body robot arm to act according to the driving variation obtained by the pre-trained neural network module and the driving compensation obtained by the online-trained neural network module.
8. An electronic device comprising a memory and a processor and computer instructions stored on the memory and executed on the processor, the computer instructions, when executed by the processor, performing the control method of any one of claims 1 to 6.
9. A robot arm system comprising a base, a continuum robot arm having one end fixedly attached to said base, a drive cable in said continuum robot arm, a drive motor for said drive cable, and a position sensor at the end of said continuum robot arm, further comprising the control system of claim 7.
CN202211280925.7A 2022-10-19 2022-10-19 Control method and system for tail end position of continuous body robot arm and robot arm system Pending CN115609585A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211280925.7A CN115609585A (en) 2022-10-19 2022-10-19 Control method and system for tail end position of continuous body robot arm and robot arm system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211280925.7A CN115609585A (en) 2022-10-19 2022-10-19 Control method and system for tail end position of continuous body robot arm and robot arm system

Publications (1)

Publication Number Publication Date
CN115609585A true CN115609585A (en) 2023-01-17

Family

ID=84863717

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211280925.7A Pending CN115609585A (en) 2022-10-19 2022-10-19 Control method and system for tail end position of continuous body robot arm and robot arm system

Country Status (1)

Country Link
CN (1) CN115609585A (en)

Similar Documents

Publication Publication Date Title
George Thuruthel et al. Learning closed loop kinematic controllers for continuum manipulators in unstructured environments
US11235461B2 (en) Controller and machine learning device
Lane et al. AMADEUS: advanced manipulation for deep underwater sampling
CN110000787B (en) Control method of super-redundant mechanical arm
Jiang et al. A two-level approach for solving the inverse kinematics of an extensible soft arm considering viscoelastic behavior
Clements et al. Three-dimensional contact imaging with an actuated whisker
Zhao et al. Shape estimation and control of a soft continuum robot under external payloads
EP2105263A2 (en) Real time control method and device for robots in virtual projection
KR20220155921A (en) Method for controlling a robot device
CN112959326B (en) Method and device for solving positive kinematics of robot, readable storage medium and robot
Zhang et al. Calibration of geometric parameters and error compensation of non-geometric parameters for cable-driven parallel robots
Zeller et al. Motion planning of a pneumatic robot using a neural network
CN115609585A (en) Control method and system for tail end position of continuous body robot arm and robot arm system
CN113910218A (en) Robot calibration method and device based on kinematics and deep neural network fusion
Yao et al. Design optimization of soft robotic fingers biologically inspired by the fin ray effect with intrinsic force sensing
CN113650014B (en) Redundant mechanical arm tracking control method based on echo state network
Li et al. A 3D printed variable cross-section pneumatic soft manipulator with high-precision positioning capability: Design and control implementation
Chen et al. Data models applied to soft robot modeling and control: A review
EP2359989B1 (en) Robot control with bootstrapping inverse kinematics
Pajak et al. Planning of a point to point collision-free trajectory for mobile manipulators
Visentin et al. A new exploration strategy for soft robots based on proprioception
CN115847430B (en) Model-free prediction full-position control method and system for soft mechanical arm
Marquez et al. Hardware-in-the-Loop Soft Robotic Testing Framework using an Actor-Critic Deep Reinforcement Learning Algorithm
US20240054393A1 (en) Learning Device, Learning Method, Recording Medium Storing Learning Program, Control Program, Control Device, Control Method, and Recording Medium Storing Control Program
US20230311303A1 (en) Method and apparatus for robotic arm navigation

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