CN-121973228-A - Robot control method based on active disturbance observation and phase coupling control

CN121973228ACN 121973228 ACN121973228 ACN 121973228ACN-121973228-A

Abstract

The invention discloses a robot control method based on active disturbance observation and phase coupling control, which comprises the steps of collecting current state quantity of a robot in real time, constructing a joint space dynamics model, generating state estimation and disturbance estimation in the joint space dynamics model, mapping phases of multi-degree-of-freedom actions to desired tracks by adopting a phase parameterization method, designing closed-loop track tracking and disturbance compensation control laws by combining disturbance estimation and a sliding manifold mechanism, designing a continuously-conductive robot auxiliary horizontal function by introducing position error transformation, analyzing stability by adopting a Lyapunov method, substituting the closed-loop track tracking and disturbance compensation control laws and a coupling gain self-adaption law into time derivatives of candidate Lyapunov functions, and completing the control of the robot. The invention can keep smooth and continuous control output, avoid abrupt change of instructions or unstable posture, maintain higher action success rate and landing stability when the environment changes, and effectively reduce the peak value of landing impact load.

Inventors

ZHANG MIN
Yan Zizhao
HE MIN
LUO SHIJUN
HUANG GUIYI
DAI WEIJIAN
YI XIANGYUAN
ZHONG MAOHUA
YE ZHENHUA

Assignees

贺州学院

Dates

Publication Date: 20260505
Application Date: 20260320

Claims (10)

1. The robot control method based on active disturbance observation and phase coupling control is characterized by comprising the following steps of: s1, collecting current state quantity of a robot in real time, and constructing a joint space dynamics model; s2, generating state estimation and disturbance estimation in a joint space dynamics model based on a component type extended state observer; s3, mapping the phase of the multi-degree-of-freedom action to a desired track by adopting a phase parameterization method, and designing a closed-loop track tracking and disturbance compensation control law by combining disturbance estimation and a sliding manifold mechanism; S4, introducing position error transformation to design a continuous and conductive robot auxiliary level function in a closed-loop track tracking and disturbance compensation control law in order to ensure that the position constraint does not conflict with an expected track; S5, analyzing stability by adopting a Lyapunov method, substituting a closed-loop track tracking and disturbance compensation control law and a coupling gain self-adaptive law into the time derivative of the candidate Lyapunov function, and completing the control of the robot.
2. The method for controlling a robot based on active disturbance observation and phase coupling control according to claim 1, wherein in S1, the current state quantity of the robot includes a joint position vector, a joint velocity vector, and a joint acceleration vector; the expression of the joint space dynamics model is specifically: In the formula, Is a matrix of inertia which is a matrix of inertia, For a Kelvin/centrifuge matrix, such that In the form of an antisymmetric matrix, As the term of gravity is used, For the actuator to output a torque/force vector, As a composite term of unknown external disturbances and modeling errors, As the joint acceleration vector, the motion vector, As a vector of the velocity of the joint, Is a joint position vector.
3. The method of claim 2, wherein in S2, the observer states are defined for the i-th degree of freedom by a component-wise extended state observer, and the state estimate in the joint space dynamics model is generated, including the position estimate Velocity estimation Disturbance estimation ; In the formula, For an inertia scale for normalization or diagonal approximation, To approximate the components of the actuator output torque/force vector, The coriolis force for the i-th degree of freedom approximates the component of the centrifugal force term, To approximate the component of the gravitational term for the i-th degree of freedom, 、 And For the component extended state observer gain, For the derivative of the i-th degree of freedom position estimate, The derivative of the velocity estimate for the ith degree of freedom, The derivative estimated for the i-th degree of freedom disturbance, The actual joint position for the i-th degree of freedom; disturbance estimation in joint space dynamics model The expression of (2) is specifically: In the formula, For the i-th degree of freedom disturbance estimation, N is the number of degrees of freedom, In order to transpose the symbol, 。
4. The method for controlling a robot based on active disturbance observation and phase coupling control according to claim 3, wherein in S3, the method for mapping the phase to the desired trajectory by using the phase parameterization method is specifically: introducing phase to degree of freedom, defining phase to desired trajectory mapping Is represented by the expression: In the formula, In order to be in the off-set position, As the amplitude coefficient of the vibration, As a basis function of phase to position, For the ith phase, define a phase rate The following coupled vibrator model gives: In the formula, The natural frequency of the i-th degree of freedom, To couple matrix elements, by The time t is integrated to obtain, In order for the phase to be adaptive in gain, For the j-th phase of the phase, Is the phase error; In the formula, Planning a given desired phase for a mission; the calculation is carried out through a coupling gain self-adaptive law, and the expression of the coupling gain self-adaptive law is specifically as follows: In the formula, In order to couple the time derivative of the gain, And In order for the rate of learning to be high, In order for the coefficient of restitution to be stable, For coupling a priori values.
5. The method for controlling a robot based on active disturbance observation and phase coupling control according to claim 4, wherein in S3, the expression of the closed-loop trajectory tracking and disturbance compensation control law is specifically: In the formula, The system is characterized in that the system is a feedback gain matrix coefficient, is used for driving the system state to converge to a sliding manifold, s is a second sliding manifold and is used for embedding the position error transformation amount and the constraint characteristic display thereof into a control input; to buffer the shot size for the impact dedicated to the landing stage, In order for the joint position vector to be desired, In order for the joint velocity vector to be desired, Is the desired joint acceleration vector.
6. The method for controlling a robot based on active disturbance observer and phase-coupled control according to claim 5, wherein a floor impact buffering strategy is used to calculate an impact buffering injection amount dedicated to a floor stage The expression is specifically as follows: In the formula, The relative velocity or contact velocity vector is estimated for the joint at the moment of contact, In order to make the opposite angle positive, Is a scale factor.
7. The method for controlling a robot based on active disturbance observation and phase coupling control according to claim 6, wherein in S3, the method for calculating the second slip manifold is specifically as follows: Constructing a first slip manifold based on position error conversion Constructing a second sliding manifold by combining the auxiliary weights ; In the formula, For the joint position tracking error vector, , For the joint velocity tracking error vector, , As a position-velocity coupling matrix, , In order to perform the diagonalization operation, To constitute a first constant of the matrix coefficients, N constants that constitute matrix coefficients; In the formula, In order to refer to the secure location vector, And generating diagonal auxiliary weight moment for the task performance index.
8. The method for controlling a robot based on active disturbance observation and phase coupling control according to claim 7, wherein in S4, the method for introducing position error transformation specifically comprises: defining tracking errors, and calculating position error conversion quantity through a position constraint conversion function; In the formula, For the i-th free position error conversion amount, For the maximum safety deviation allowed for the ith degree of freedom, Is the position tracking error of the i-th degree of freedom.
9. The method for controlling a robot based on active disturbance observation and phase coupling control according to claim 8, wherein in S4, the method for designing the continuously-conductive robot assistance level function is specifically as follows: Defining a task performance function, and calculating diagonal auxiliary weight moment through the task performance function , , A task performance function for the ith degree of freedom; In the formula, For the measured or estimated human interaction force/moment components, And Is based on preset weight Design of continuously guided robot-assisted level functions The expression is specifically as follows: In the formula, As a boundary of the dead zone, As the boundary of the saturation region, Is the slope coefficient of the hyperbolic tangent function, is used for adjusting the steepness degree of the function in the transition zone, Is the center offset of the hyperbolic tangent function and is used for determining the excessive center position of the function.
10. The method for controlling a robot based on active disturbance observation and phase coupling control according to claim 9, wherein in S5, the lyapunov method is specifically: defining tracking error, and using candidate Lyapunov function The expression of (2) is specifically: In the formula, In order to couple the target value or the desired value, For the positional potential energy gain matrix coefficients in the lyapunov function, For candidate Lyapunov functions Regarding time derivative, the closed-loop trajectory tracking and disturbance compensation control law and the coupling gain adaptive law are substituted into Is obtained under conditions of satisfying the dynamics assumption and reasonable gain selection: In the formula, As candidate lyapunov function For evaluating the rate of change of energy and the stability of the system.

Description

Robot control method based on active disturbance observation and phase coupling control Technical Field The invention belongs to the technical field of intelligent robot motion control, and particularly relates to a robot control method based on active disturbance observation and phase coupling control. Background High-speed motion robots are gaining more and more attention in applications such as complex terrain exploration, disaster relief, and motion interaction. Jumping is an important means for realizing crossing obstacles and high dynamic terrain adaptability as a movement mode breaking through the limit of continuous ground contact. However, most robots still rely mainly on steady walking or rolling movements, and have limited ability to face high steps, spacing obstacles or flexible terrain, making continuous jumping and high stability movements difficult to achieve. Existing robot jump control methods mostly rely on predefined motion trajectories or on off-line solution strategies based on kinetic models. Under ideal conditions, such control means enable a complete jump motion. However, in practical application, factors such as uncertainty of model parameters, flexible deformation of a mechanism, performance difference of a driver, and ground contact conditions can cause degradation of control precision, so that a motion track deviates from expectations. Because this type of approach lacks real-time feedback capability for environmental changes and state disturbances, it is generally only suitable for repetitive or relatively static scenes during a jump. When external conditions change, the control strategies cannot adjust the track or output moment in time, and particularly larger impact load and attitude deviation are easier to generate in the landing stage, so that the stability of the robot jumping process and the task execution success rate are reduced. In order to make up for the defect of the fixed track control method in the aspect of adaptability, some researches introduce a real-time adjustment mechanism based on sensor feedback, so that a control system can dynamically adjust control output according to error change. However, such error response based control strategies are inherently limited by the response hysteresis of the feedback loop. When the change speed of the motion state variables such as the joint angle, the angular speed or the system centroid track is high in the jumping process, or the amplitude of external disturbance (such as abrupt change of ground reaction force, change of contact condition and the like) is high, the compensation mechanism cannot complete effective error correction in a single control period, so that control command change discontinuity, output torque fluctuation and gesture control oscillation are caused. The above problems further reduce the continuity of the jump motion and the stability of the control process, thereby affecting the controllability and reliability of the system in high dynamic scenarios. In addition, a jump control method based on a prediction model is proposed, and the stability and the landing consistency of jump actions are improved by planning a motion sequence in advance and combining a constraint optimization algorithm. Although such methods exhibit higher performance in theoretical modeling and ideal simulation environments, their control effects are highly dependent on accurate system model parameters and strong computational resource support. In actual operation, the prediction control method is difficult to meet the real-time requirement due to calculation delay, model error accumulation and environmental uncertainty. When high dynamic tasks such as continuous jump, change of friction coefficient across the ground or rapid posture adjustment are performed, problems such as control response lag, track deviation increase or stability reduction in the landing stage are easy to occur, so that the control performance of the robot cannot continuously keep the expected level. In summary, the conventional jump control method has the common performance bottleneck in practical application that the control mode depending on a predefined track is difficult to adapt to dynamic environment change, the compensation strategy based on error feedback is limited by feedback delay, continuous and stable control output is difficult to realize in the high dynamic transition process, and the control mode based on a prediction model has theoretical performance advantages, but has obvious dependence on accurate model identification and high computing resources, and is difficult to meet real-time control requirements. With the increase of the complexity of jumping actions, such as continuous transition, variable friction condition contact, rapid gesture reconfiguration and other task scenes, the problems are further amplified, and the problems are represented by the phenomena of increased track deviation, enhanced landing impact, reduced output ene