CN-121973212-A - Mechanical arm self-adaptive variable-impedance constant-force control system

CN121973212ACN 121973212 ACN121973212 ACN 121973212ACN-121973212-A

Abstract

The invention discloses a mechanical arm self-adaptive variable-impedance constant-force control system which comprises an inverse kinematics unit, a fractional order active disturbance rejection controller unit, a mechanical arm and a self-adaptive variable-impedance controller unit, wherein the input of the system is the expected position of the tail end and the expected contact force. The invention obviously reduces force overshoot, the steady state error approaches zero, and improves the force tracking precision and robustness.

Inventors

WU QIBIAO
Cao Zhexian
ZHONG RUJIE
LIU XINMENG

Assignees

上海睿镜通医疗科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260204

Claims (10)

1.A mechanical arm self-adaptive variable-impedance constant-force control system is characterized by comprising an inverse kinematics unit, a fractional order active disturbance rejection controller unit, a mechanical arm and a self-adaptive variable-impedance controller unit, The inputs to the system are the tip desired position X d and the desired contact force F d , The end desired position X d and the position correction output by the adaptive variable impedance controller unit Superimposed as an end reference trajectory X r , The inverse kinematics unit receives X r and outputs the desired position q of the joint space to a fractional order active disturbance rejection controller unit, The fractional order active disturbance rejection controller unit receives q and an actual joint position q d output by a joint angle sensor of the mechanical arm and outputs a control quantity u to the mechanical arm, The mechanical arm receives u to move, the joint angle sensor of the mechanical arm outputs the actual joint angle q d to the fractional order active disturbance rejection controller unit of the mechanical arm, The force sensor at the tail end of the mechanical arm collects the actual contact force F a at the tail end of the mechanical arm, the expected contact force F d is used for subtracting F a and taking the absolute value to obtain the force error F, The adaptive variable impedance controller unit receives F and outputs a position correction amount 。
2. The adaptive variable impedance constant force control system according to claim 1, wherein said fractional order active-disturbance-rejection controller unit comprises a tracking controller module, a fractional order state feedback control law module, a nonlinear extended state observer module, The tracking controller module receives the desired position q of the joint space, outputs a tracking signal q 1 and a differential signal q 2 , The fractional state feedback control law module receives q 1 、 q 2 and z 1、 z 2 ,z 1 output by the nonlinear extended state observer module is state tracking of the actual speed q 1 , z 2 is state tracking of the differential signal q 2 , the fractional state feedback control law module outputs intermediate quantity u 0 ,u 0 , z 3 /b 0 is overlapped and then is output to the mechanical arm as control quantity u, z 3 is unknown interference estimated value of the system output by the nonlinear extended state observer module, b 0 is compensation factor, The actual joint position q d output by the joint angle sensor of the mechanical arm is overlapped with b 0 u as input of the nonlinear extended state observer module, and the nonlinear extended state observer module outputs z 1 、z 2 、z 3 .
3. The robot arm adaptive variable impedance constant force control system according to claim 2, wherein the adaptive variable impedance controller unit comprises an impedance control module and an RBF neural network module, The RBF neural network module receives F and outputs an impedance parameter correction quantity delta B to the impedance control module, The impedance control module receives F and DeltaB and outputs a position correction amount 。
4. The adaptive variable impedance constant force control system of claim 3, wherein the system optimizes parameters using a modified artificial lemming optimization algorithm IALA, the parameters including a proportional coefficient K p , an integral coefficient K i , a differential coefficient K d , a center c and a width b of neurons of an hidden layer of the neural network module of the fractional state feedback control law module.
5. The robotic arm adaptive variable impedance constant force control system according to claim 2, wherein the nonlinear extended state observer module employs a nonlinear smoothing function 。
6. The robot-adaptive variable-impedance constant-force control system according to claim 2, wherein the fractional-order active-disturbance-rejection controller unit employs a Caputo-type fractional-order derivative: Wherein, the Is a function of the Gamma of the light source, , , 。
7. The robot arm adaptive variable impedance constant force control system according to claim 3, wherein said RBF neural network module structure comprises: The input layer inputs a force error F as a variable; An underlying layer comprising 5 neurons A gaussian kernel function is used as the activation function: Wherein, the As the center of the neuron element, Is a width parameter; The convolution layer, RBF neural network module output carries out linear combination through sliding window mode, forms local connection characteristic: Wherein, the For sharing convolution kernel weights; an activation layer by convolving the result Input into a fusion type activation function, the nonlinear expression capacity of the network is enhanced, and the output is that Wherein, the Respectively controlling the weights and the offset of the two types of activation functions as the learnable parameters; An output layer for obtaining the damping coefficient of the final output by weighting and summing the output of the activation layer Wherein, the In order to output the layer weight parameters, As a damping adjustment in the impedance controller, the drive system achieves the desired dynamic compliance behavior.
8. The robot arm adaptive variable impedance constant force control system according to claim 7, wherein the parameter adjustment method of the RBF neural network module is as follows: the parameters of the neural network are adjusted according to the performance index function E (t), In the formula, Is the first The desired force at the moment in time, Is the first The contact force at the moment in time is, Is the first Force error at time; and minimizing the performance index function by a gradient descent method, reversely propagating error signals layer by layer, and updating each parameter.
9. The adaptive variable impedance constant force control system according to claim 4, wherein optimizing parameters with the improved manual lemming optimization algorithm IALA comprises a gradient-guided greedy acceleration mechanism, a Gaussian-disturbance-based adaptive local search, a diversity-enhancement mechanism, and an adaptive dynamic adjustment strategy.
10. The mechanical arm adaptive variable impedance constant force control system of claim 9, wherein optimizing parameters with the modified artificial lemming optimization algorithm IALA further comprises coordinating a fractional order state feedback control law module and an RBF neural network module parameter optimization process using a joint fitness function, the joint fitness function being: Wherein: Representing an adaptive function of the inner loop of the position, Representing the adaptive function of the force control outer ring, The weight coefficients of the two are respectively used for adjusting optimization bias.

Description

Mechanical arm self-adaptive variable-impedance constant-force control system Technical Field The invention relates to the field of robots, in particular to a self-adaptive variable-impedance constant-force control system for a mechanical arm. Background With the continuous development of robot control technology, the robot control technology is widely applied to a plurality of key fields such as industrial manufacture, medical operation, space detection and the like, and gradually becomes an important support technology for realizing automatic and high-precision operation. In an actual operation scene, the robot is often in physical contact with an uncertain and dynamic environment, such as polishing, grabbing, operation and other tasks, which puts higher demands on force control precision, flexibility and environment adaptability. In order to realize safe and stable interaction between the robot and the environment, constant force control gradually becomes a research hot spot, and the current common constant force control method mainly comprises impedance control and admittance control. The impedance control method has the advantages of coordinating force-position relation, adjusting flexibility and certain anti-interference capability, and is dominant in constant force control. However, the conventional impedance control still has many challenges when facing complex operation scenarios, such as severe environmental stiffness variation and fixed control parameters, which results in difficult self-adaptation of the system, severe overshoot and oscillation of force response, and affects control accuracy and safety. Accordingly, those skilled in the art have been directed to developing a novel mechanical arm adaptive variable impedance constant force control system. Disclosure of Invention In view of the above-mentioned drawbacks of the prior art, the technical problems to be solved by the present invention are: 1. The traditional impedance control parameters are fixed, so that the environment rigidity change is difficult to adapt, and the force is overlarge and the vibration is serious; 2. The traditional RBF neural network has low feature extraction efficiency and lag in parameter updating; 3. The parameter setting of the controller depends on manual trial and error, so that multi-objective collaborative optimization is difficult to realize; 4. Conventional ADRCs are prone to high frequency oscillations in highly nonlinear systems. In order to achieve the aim, the invention provides a mechanical arm self-adaptive variable impedance constant force control system which comprises an inverse kinematics unit, a fractional order active disturbance rejection controller unit, a mechanical arm and a self-adaptive variable impedance controller unit, The inputs to the system are the tip desired position X d and the desired contact force F d, End desired position X d and position correction output by adaptive variable impedance controller unitSuperimposed as an end reference trajectory X r, The inverse kinematics unit receives X r and outputs the desired position q of the joint space to a fractional order active-disturbance-rejection controller unit, The fractional order active disturbance rejection controller unit receives q and an actual joint position q d output by a joint angle sensor of the mechanical arm, and outputs a control quantity u to the mechanical arm, The mechanical arm receives u to move, the joint angle sensor of the mechanical arm outputs the actual joint angle q d to the fractional order active disturbance rejection controller unit of the mechanical arm, The force sensor at the tail end of the mechanical arm collects the actual contact force F a at the tail end of the mechanical arm, the expected contact force F d is used for subtracting F a and taking the absolute value to obtain the force error F, The adaptive variable impedance controller unit receives F and outputs a position correction amount。 Further, the fractional order active disturbance rejection controller unit comprises a tracking controller module, a fractional order state feedback control law module and a nonlinear extended state observer module, The tracking controller module receives the desired position q of the joint space, outputs a tracking signal q 1 and a differential signal q 2, The fractional state feedback control law module receives the state tracking of the actual speed q 1 by the z 1、z2,z1 output by the q 1、 q2 and the nonlinear extended state observer module, the state tracking of the differential signal q 2 by the z 2, the output intermediate quantity u 0,u0 of the fractional state feedback control law module is overlapped with the z 3/b0 and then is output to the mechanical arm as a control quantity u, the unknown interference estimated value of the system output by the nonlinear extended state observer module by the z 3 is b 0 is a compensation factor, The actual joint position q d output by the joint angle sensor of