Robot control method based on active disturbance observation and phase coupling control

Abstract

The invention discloses a robot control method based on active disturbance observation and phase coupling control, and the method comprises the steps: collecting the current state quantity of a robot in real time, and constructing a joint space dynamic model; generating state estimation and disturbance estimation in the joint space dynamic model; a phase parameterization method is adopted to map the phase of the multi-degree-of-freedom action to an expected trajectory, and a closed-loop trajectory tracking and disturbance compensation control law is designed in combination with disturbance estimation and a sliding manifold mechanism; according to the method, a position error transformation design continuously-derivable robot auxiliary level function is introduced, the stability is analyzed by adopting a Lyapunov method, and a closed-loop trajectory tracking and disturbance compensation control law and a coupling gain adaptive law are substituted into a time derivative of a candidate Lyapunov function to complete the control of the robot. The control output can be kept smooth and continuous, instruction sudden change or attitude instability is avoided, high action success rate and landing stability can still be maintained when the environment changes, and the landing impact load peak value is effectively reduced.

Description

Technical Field

[0001] This invention belongs to the field of intelligent robot motion control technology, specifically relating to a robot control method based on active disturbance observation and phase coupling control. Background Technology

[0002] High-speed robots are gaining increasing attention in applications such as complex terrain exploration, disaster relief, and motion interaction. Jumping, as a mode of locomotion that overcomes the limitations of continuous ground contact, is an important means of achieving obstacle crossing and adaptability to highly dynamic terrain. However, most robots currently rely primarily on steady-state walking or rolling for movement, and their capabilities are limited when facing high steps, intermittent obstacles, or flexible terrain, making it difficult to achieve continuous jumps and highly stable movements.

[0003] Most existing robot jumping control methods rely on predefined motion trajectories or offline solution strategies based on dynamic models. Under ideal conditions, such control methods can achieve complete jumping movements. However, in practical applications, factors such as model parameter uncertainties, mechanism flexibility deformation, differences in actuator performance, and landing contact conditions can lead to a decrease in control accuracy and cause the motion trajectory to deviate from the expected direction. Because these methods lack real-time feedback capabilities for environmental changes and state disturbances, they are typically only suitable for repetitive or relatively static scenarios during jumping. When external conditions change, these control strategies cannot adjust the trajectory or output torque in a timely manner, especially during the landing phase, which is more prone to generating large impact loads and attitude deviations, thereby reducing the stability of the robot's jumping process and the success rate of task execution.

[0004] To compensate for the limitations of fixed-track control methods in terms of adaptability, some studies have introduced real-time adjustment mechanisms based on sensor feedback, enabling the control system to dynamically adjust the control output according to error changes. However, this type of error-response-based control strategy is inherently limited by the response lag of the feedback loop. When motion state variables such as joint angles, angular velocities, or the trajectory of the system's center of mass change rapidly during a jump, or when external disturbances (such as sudden changes in ground reaction force or contact conditions) are significant, the compensation mechanism often cannot complete effective error correction within a single control cycle. This leads to discontinuous changes in control commands, fluctuations in output torque, and oscillations in attitude control. These problems further reduce the continuity of the jump action and the stability of the control process, thus affecting the controllability and reliability of the system in highly dynamic scenarios.

[0005] Other studies have proposed jump control methods based on predictive models, which improve the stability of jump actions and landing consistency by pre-planning motion sequences and combining them with constraint optimization algorithms. Although such methods show high performance in theoretical modeling and ideal simulation environments, their control effectiveness is highly dependent on accurate system model parameters and strong computational resources. In actual operation, due to computational latency, model error accumulation, and the difficulty in fully modeling environmental uncertainties, these predictive control methods often fail to meet real-time requirements. When performing high-dynamic tasks such as continuous jumps, crossing changes in ground friction coefficients, or rapid attitude adjustments, problems such as control response lag, increased trajectory deviation, or decreased stability during landing are prone to occur, causing the robot's control performance to fail to maintain the expected level.

[0006] In summary, existing jump control methods generally suffer from performance bottlenecks in practical applications: control methods relying on predefined trajectories struggle to adapt to dynamic environmental changes; error feedback-based compensation strategies are limited by feedback delays, making it difficult to achieve continuous and stable control outputs during highly dynamic transitions; while predictive model-based control offers theoretical performance advantages, it is heavily reliant on accurate model identification and high computational resources, making it difficult to meet real-time control requirements. As the complexity of jump actions increases, such as in scenarios involving continuous transitions, variable friction contact, and rapid attitude reconfiguration, these problems are amplified, manifesting as increased trajectory deviation, enhanced landing impact, reduced output energy utilization efficiency, and decreased attitude control stability. Therefore, a jump control method capable of real-time disturbance perception, adaptive cross-stage coordinated control, and continuous motion generation throughout the entire dynamic behavior cycle is still lacking. Developing a jump control strategy with real-time adjustment capabilities, high dynamic response performance, and environmental adaptability is of significant technical importance and application value for improving the motion reliability and task execution capabilities of robots in complex environments. Summary of the Invention

[0007] To address the aforementioned shortcomings in existing technologies, the robot control method based on active perturbation observation and phase coupling control provided by this invention solves the problems of discontinuous control mode switching, asynchronous joint movement rhythm, and excessive transient load during landing in existing robots during jumping, crossing, and rapid posture adjustment.

[0008] To achieve the aforementioned objectives, the present invention employs the following technical solution: a robot control method based on active perturbation observation and phase coupling control, comprising the following steps: S1. Real-time acquisition of the robot's current state variables to construct a joint space dynamics model; S2. Based on the component-based extended state observer, generate state estimates and disturbance estimates in the joint space dynamics model; S3. The phase parameterization method is used to map the phase of the multi-degree-of-freedom motion to the desired trajectory, and the closed-loop trajectory tracking and disturbance compensation control law is designed by combining disturbance estimation and sliding manifold mechanism. S4. To ensure that the position constraints and the desired trajectory do not conflict, a continuously differentiable robot-aided level function is designed by introducing position error transformation in the closed-loop trajectory tracking and disturbance compensation control law. S5. Stability is analyzed using the Lyapunov method. The closed-loop trajectory tracking and disturbance compensation control law and the coupling gain adaptive law are substituted into the time derivative of the candidate Lyapunov function to complete the robot control.

[0009] Furthermore: In S1, the robot's current state variables include joint position vector, joint velocity vector, and joint acceleration vector; The specific expression for the joint space dynamics model is as follows: In the formula, The inertia matrix, For the Coriolis / eccentric matrix, such that For antisymmetric matrices, For gravity, To output torque / force vector to the actuator, This is a composite term of unknown external disturbances and modeling errors. The joint acceleration vector, For joint velocity vectors, This is the joint position vector.

[0010] Furthermore: In S2, the component-based extended state observer is used to observe the first... i Degrees of freedom define the observer state and generate state estimates in the joint space dynamics model, including position estimates. Speed ​​estimation and disturbance estimation ; In the formula, For use as a scale of inertia for normalization or diagonal approximation, To approximate the components of the actuator output torque / force vector, For the first i The Coriolis force and the components of the centrifugal force term are approximately equal in degree of freedom. For the first i The components of the gravity term in the degrees of freedom are approximated. , and For component-based extended state observer gain, For the first i The derivative of the position estimation of degrees of freedom, For the first i The derivative of the velocity estimation for degrees of freedom, For the first i The derivative of the degree-of-freedom perturbation estimate, For the first i The actual joint positions of the degrees of freedom; Perturbation estimation in joint space dynamics model The specific expression is: In the formula, For the first i Degrees of freedom perturbation estimation , n For the number of degrees of freedom, It is the transpose symbol. .

[0011] Furthermore: In S3, the method of mapping the phase to the desired trajectory using phase parameterization is specifically as follows: Introduce phase to the degrees of freedom and define the phase-to-desired trajectory mapping. The expression: In the formula, For offset position, The amplitude coefficient, Let be the basis functions from phase to position. For the first i Phase, specifying the phase rate The following coupled oscillator model is given: In the formula, No. i The natural frequency of degrees of freedom For the elements of the coupling matrix, by Regarding time t By integration, For phase adaptive gain, For the first j Phase, This is the phase error; In the formula, The desired phase is given for task planning; The calculation is performed using the adaptive coupling gain law, the specific expression of which is: In the formula, The time derivative of the coupling gain. and For learning rate, This is the stabilization coefficient. These are coupling prior values.

[0012] Furthermore: In S3, the specific expression for the closed-loop trajectory tracking and disturbance compensation control law is as follows: In the formula, These are the feedback gain matrix coefficients, used to drive the system state to converge to the sliding manifold. s The second sliding manifold is used to embed the position error transformation and its constraint characteristics into the control input. This is the amount of shock buffer injection specifically designed for the landing phase. Let the desired joint position vector be... Let the desired joint velocity vector be... Let be the desired joint acceleration vector.

[0013] Furthermore: A landing impact buffering strategy is employed to calculate the impact buffer injection volume specifically for the landing phase. Its expression is as follows: In the formula, Estimate the relative velocity or contact velocity vector of the joint at the moment of contact. It is diagonally positive definite. is the scaling factor.

[0014] Furthermore: In S3, the method for calculating the second sliding manifold is as follows: Construct the first sliding manifold based on the position error transformation. Then, by combining auxiliary weights, a second sliding manifold is constructed. ; In the formula, This is the joint position tracking error vector. , This is the joint velocity tracking error vector. , This is the position-velocity coupling matrix. , For diagonalization operation, It is the first constant that makes up the coefficients of the matrix. The first of the coefficients of the matrix n constant; In the formula, For reference safe position vector, Diagonal auxiliary weight moments generated for task performance metrics.

[0015] Furthermore, in S4, the method for introducing position error transformation is as follows: Define the tracking error and calculate the position error transformation amount using the position constraint transformation function; In the formula, For the first i Free position error conversion amount, For the first i The maximum safety deviation allowed for degrees of freedom For the first i Position tracking error for degrees of freedom.

[0016] Furthermore: In S4, the specific method for designing a continuously differentiable robot-aided level function is as follows: Define a task performance function, and calculate the diagonal auxiliary weight moments using the task performance function. , , For the first i Task performance function with degrees of freedom; In the formula, To measure or estimate the force / torque components of human-computer interaction, and For the preset weights, based on Design a continuously differentiable robot-aided level function. Its expression is as follows: In the formula, The boundary of the dead zone This is the boundary of the saturation region. This is the slope coefficient of the hyperbolic tangent function, used to adjust the steepness of the function in the transition region. This is the center offset of the hyperbolic tangent function, used to determine the center position of the function's transition.

[0017] Furthermore: In S5, the Lyapunov method specifically refers to: Define the tracking error and use the candidate Lyapunov function. The specific expression is: In the formula, To couple the target value or desired value, These are the coefficients of the position potential gain matrix in the Lyapunov function. For candidate Lyapunov functions Regarding the time derivative, substituting the closed-loop trajectory tracking and disturbance compensation control law and the coupling gain adaptive law into... The expression for , under the conditions of satisfying the dynamic assumptions and reasonable gain selection, can be obtained as follows: In the formula, Candidate Lyapunov functions The time derivative is used to evaluate the rate of energy change and stability of the system.

[0018] The beneficial effects of this invention are as follows: (1) This invention provides a robot control method based on active disturbance observation and phase coupling control. By estimating external disturbances and model errors in real time and combining a phase-driven desired trajectory generation mechanism, continuous and adjustable motion control is achieved throughout the jumping process. During the motion, the system dynamically adjusts the coupling gain according to the changes in joint state to maintain the motion coordination of multi-degree-of-freedom joints, and performs linkage control of each stage of take-off, take-off, landing and recovery through a unified control law. When switching between different control modes (including autonomous control and human-machine collaborative control), this method can maintain smooth and continuous control output, avoid sudden changes in commands or attitude instability, and thus maintain a high success rate of action and landing stability even when the environment and contact conditions change. At the same time, it effectively reduces the peak value of landing impact load, and solves the problems of discontinuous control mode switching, asynchronous joint movement rhythm and excessive transient load during landing in existing robots during jumping, crossing and rapid attitude adjustment.

[0019] (2) This invention introduces safety constraints and disturbance compensation into the control law through position error conversion and sliding manifold mechanism, ensuring that position constraints and phase-driven trajectory do not conflict, thereby improving the robot's motion stability, action continuity and landing safety in different environments.

[0020] (3) The present invention adopts a phase coupling and adaptive phase adjustment method. The phase coupling is realized through a coupled oscillator model. The coupled oscillator model realizes the motion coordination of multiple joints by introducing a phase difference sine term. The adaptive phase adjustment method is realized through a coupling gain adaptive law. The coupling gain adaptive law dynamically adjusts the coupling gain of the phase in real time, so that the multi-degree-of-freedom joints maintain phase coordination in each stage of take-off, take-off, landing and recovery, avoiding discontinuous movement, shaking or instability caused by traditional hard switching, and ensuring the controllability of continuous motion.

[0021] (4) The present invention realizes overall motion planning and execution based on a unified control framework, enabling the robot to safely and continuously complete complex tasks such as obstacle crossing, jumping, dynamic operation and rehabilitation assistance. Attached Figure Description

[0022] Figure 1 This is a flowchart of a robot control method based on active perturbation observation and phase coupling control. Detailed Implementation

[0023] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0024] like Figure 1 As shown, in one embodiment of the present invention, a robot control method based on active perturbation observation and phase coupling control includes the following steps: S1. Real-time acquisition of the robot's current state variables to construct a joint space dynamics model; S2. Based on the component extended state observer (ESO), generate state estimates and disturbance estimates in the joint space dynamics model; S3. The phase parameterization method is used to map the phase of the multi-degree-of-freedom motion to the desired trajectory, and the closed-loop trajectory tracking and disturbance compensation control law is designed by combining disturbance estimation and sliding manifold mechanism. S4. To ensure that the position constraints and the desired trajectory do not conflict, a continuously differentiable robot-aided level function is designed by introducing position error transformation in the closed-loop trajectory tracking and disturbance compensation control law to satisfy the continuous robot control and stability guarantee under position constraints. S5. Stability is analyzed using the Lyapunov method. The closed-loop trajectory tracking and disturbance compensation control law and the coupling gain adaptive law are substituted into the time derivative of the candidate Lyapunov function to complete the robot control.

[0025] In S1, the robot's current state variables include joint position vector, joint velocity vector, and joint acceleration vector; The specific expression for the joint space dynamics model is as follows: In the formula, The inertia matrix is ​​symmetric positive definite. For the Coriolis / eccentric matrix, such that For antisymmetric matrices, This is a gravity term (or other location-related term). To output torque / force vector to the actuator, This is a composite term of unknown external disturbances and modeling errors. The joint acceleration vector, For joint velocity vectors, This is the joint position vector.

[0026] In S2, the component-based extended state observer is used to observe the first... i Degrees of freedom define the observer state and generate state estimates in the joint space dynamics model, including position estimates. Speed ​​estimation and disturbance estimation ; In the formula, For use as a scale of inertia for normalization or diagonal approximation, , To approximate the components of the actuator output torque / force vector, For the first i The Coriolis force and centrifugal force components of the degrees of freedom are approximations; it is a simplification or approximation of the velocity-related nonlinear terms in the robot's dynamics model. For the first i The gravity component of the degrees of freedom is an approximation, representing the projection or approximation of the gravitational influence on the robot at the current joint. , and For component-based extended state observer gain, , and All are greater than 0. For the first i The derivative of the position estimation of degrees of freedom, For the first i The derivative of the velocity estimation for degrees of freedom, For the first i The derivative of the degree-of-freedom perturbation estimate, For the first i The actual joint positions of the degrees of freedom; Perturbation estimation in joint space dynamics model The specific expression is: In the formula, For the first i Degrees of freedom perturbation estimation , n For the number of degrees of freedom, It is the transpose symbol. .

[0027] In S3, the method of mapping the phase to the desired trajectory using phase parameterization is as follows: Introduce phase to the degrees of freedom and define the phase-to-desired trajectory mapping. The expression: In the formula, For offset position, The amplitude coefficient, , Let be the basis functions from phase to position. For the first i Phase, the i phase From phase rate Regarding time t The phase rate is obtained by integration. The following coupled oscillator model is given: In the formula, No. i The natural frequency of the degrees of freedom (which can be scheduled according to the action phase). For the elements of the coupling matrix, by Regarding time t Integrating, we obtain the result, which reflects the first... i With the j Interphase coupling strength, For phase adaptive gain, , For the first j Phase, This is the phase error; In the formula, The desired phase is given for task planning.

[0028] To achieve adaptive coupling under disturbances or task changes, an adaptive coupling gain law is designed. The specific expression of the adaptive coupling gain law is as follows: In the formula, The time derivative of the coupling gain. and For learning rate, , , This is the stabilization coefficient. , As a coupled prior value, this adaptive law enhances coupling when synchronization requirements increase and returns to the desired trajectory when stable, ensuring the robot's robustness.

[0029] In S3, the specific expression for the closed-loop trajectory tracking and disturbance compensation control law is as follows: In the formula, These are the feedback gain matrix coefficients, used to drive the system state to converge to the sliding manifold. s The second sliding manifold is used to embed the position error transformation and its constraint characteristics into the control input. This embodiment provides the shock buffer injection volume specifically for the landing phase. One implementation form, Let the desired joint position vector be... Let the desired joint velocity vector be... Let be the desired joint acceleration vector.

[0030] In this embodiment, , and All belong to the desired trajectory, and are mapped from phase to desired trajectory. Obtained, according to The expression utilizes the current phase The desired joint position vector is obtained through basis function mapping. For the desired joint position vector The desired joint velocity vector is obtained by taking the first derivative with respect to time. For the desired joint velocity vector Differentiation over time, or with respect to the desired joint position vector The desired joint acceleration vector is obtained by taking the second derivative with respect to time. .

[0031] In this embodiment, the following is given One specific implementation involves: to reduce the impact and oscillation at the moment of robot contact with the ground, introducing saturated damping based on estimated relative velocity during the landing phase, and employing a landing impact buffering strategy to calculate the impact buffer injection amount specifically for the landing phase. Its expression is as follows: In the formula, Estimate the relative velocity or contact velocity vector of the joint at the moment of contact. It is diagonally positive definite. The scaling factor is... Tanh is used to ensure that the injected terms are bounded and smooth.

[0032] In S3, the method for calculating the second sliding manifold is as follows: Construct the first sliding manifold based on the position error transformation. Then, by combining auxiliary weights, a second sliding manifold is constructed. ; In the formula, This is the joint position tracking error vector. , This is the joint velocity tracking error vector. , This is the position-velocity coupling matrix. , To represent a diagonal matrix with the elements within parentheses as the main diagonal elements, It is the first constant that makes up the coefficients of the matrix. The first of the coefficients of the matrix n constant; In this embodiment, the first sliding manifold Responsible for handling the robot's trajectory tracking performance and hard position safety constraints, as long as the system is... On the synovial surface, the error can be guaranteed to converge within a safe range.

[0033] In the formula, The reference safe position vector (which can be an offset or zero vector) Diagonal auxiliary weight moments generated for task performance metrics.

[0034] In this embodiment, the second sliding manifold This is the core of the "phase coupling" in this invention; it is... Based on this, diagonal auxiliary weight moments representing task performance are introduced. When the task performs poorly or is in a specific phase, through... Adjusting the synovial surface can alter the robot's compliance or tracking behavior.

[0035] In S4, the method for introducing position error transformation is as follows: Define the tracking error and calculate the position error transformation amount using the position constraint transformation function; In the formula, For the first i Free position error conversion amount, For the first i The maximum safety deviation allowed for degrees of freedom , For the first i Position tracking error for degrees of freedom; In this embodiment, the present invention uses a formula to define the constrained first... i Position tracking error of degrees of freedom Convert to the first i Free position error conversion amount ,when Approaching the set numberi Maximum safety deviation allowed for degrees of freedom hour, It will tend towards infinity. Through control... Boundedness provides a rigorous mathematical guarantee for actual error. It will never exceed the safe range, satisfying continuous robot control and stability assurance under positional constraints, thereby achieving safe constraint control of the robot.

[0036] In S4, the specific method for designing a continuously differentiable robot-aided level function is as follows: Define a task performance function, and calculate the diagonal auxiliary weight moments using the task performance function. , , For the first i Task performance function with degrees of freedom; In the formula, To measure or estimate the force / torque components of human-computer interaction, and For the preset weights, , ,based on Design a continuously differentiable robot-aided level function. It satisfies: In the formula, The boundary of the dead zone This is the boundary of the saturation region. To ensure Continuity, To indicate that the first derivative of a function is continuous, cubic splines or tanh can be used to approximate a smooth transition region. Specifically, a smooth, monotonically increasing function is defined as follows: In the formula, This is the slope coefficient of the hyperbolic tangent function, used to adjust the steepness of the function in the transition region. This is the center offset of the hyperbolic tangent function, used to determine the center position of the function's transition. It is determined by selecting... and Make In the interval The endpoint values ​​match the first derivative. The boundary of the dead zone This represents the boundary of the saturation region.

[0037] At the edge of the dead zone and saturation region boundary At this point, the transition curve constructed by the hyperbolic tangent function (tanh) must have its function value and first derivative value (slope) smoothly connected to the function definition outside the interval (usually 0, 1, or a linear segment). This ensures... It is continuous and differentiable (smooth) throughout the entire domain, avoiding abrupt changes or jitter in the control signal.

[0038] In S5, the Lyapunov method is specifically as follows: Define the tracking error and use the candidate Lyapunov function. The specific expression is: In the formula, To couple the target value or desired value, These are the coefficients of the position potential gain matrix (or stiffness matrix) in the Lyapunov function. For candidate Lyapunov functions Differentiating over time yields the candidate Lyapunov function. time derivative : For candidate Lyapunov functions Regarding the time derivative, substituting the closed-loop trajectory tracking and disturbance compensation control law and the coupling gain adaptive law into... The expression for , under the conditions of satisfying the dynamic assumptions and reasonable gain selection, can be obtained as follows: In the formula, Candidate Lyapunov functions The time derivative is used to evaluate the rate of energy change and stability of the system.

[0039] The specific process is as follows: First, for candidate Lyapunov functions Taking the derivative with respect to time, we get... The expansion of .

[0040] Secondly, the expression of the robot's joint space dynamics model. Substitution The expansion formula, and the closed-loop trajectory tracking and disturbance compensation control law. Substitution The expansion of; Finally, we utilize an important property of robot dynamics: the matrix. By utilizing the oblique symmetry to eliminate relevant nonlinear terms, the final expression is simplified and obtained. After cancellation by the control law, the final derivation is obtained. This ensures that the closed-loop error is bounded and tends to stabilize.

[0041] In the description of this invention, the above are merely preferred embodiments and are not intended to limit the scope of protection of this invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A robot control method based on active perturbation observation and phase coupling control, characterized in that, Includes the following steps: S1. Real-time acquisition of the robot's current state variables to construct a joint space dynamics model; S2. Based on the component-based extended state observer, generate state estimates and disturbance estimates in the joint space dynamics model; S3. The phase parameterization method is used to map the phase of the multi-degree-of-freedom motion to the desired trajectory, and the closed-loop trajectory tracking and disturbance compensation control law is designed by combining disturbance estimation and sliding manifold mechanism. S4. To ensure that the position constraints and the desired trajectory do not conflict, a continuously differentiable robot-aided level function is designed by introducing position error transformation in the closed-loop trajectory tracking and disturbance compensation control law. S5. Stability is analyzed using the Lyapunov method. The closed-loop trajectory tracking and disturbance compensation control law and the coupling gain adaptive law are substituted into the time derivative of the candidate Lyapunov function to complete the robot control.

2. The robot control method based on active perturbation observation and phase coupling control according to claim 1, characterized in that, In S1, the robot's current state variables include joint position vector, joint velocity vector, and joint acceleration vector; The specific expression for the joint space dynamics model is as follows: In the formula, The inertia matrix, For the Coriolis / eccentric matrix, such that For antisymmetric matrices, For gravity, To output torque / force vector to the actuator, This is a composite term of unknown external disturbances and modeling errors. The joint acceleration vector, For joint velocity vectors, This is the joint position vector.

3. The robot control method based on active perturbation observation and phase coupling control according to claim 2, characterized in that, In S2, the component-based extended state observer is used to observe the first... i Degrees of freedom define the observer state and generate state estimates in the joint space dynamics model, including position estimates. Speed ​​estimation and disturbance estimation ; In the formula, For use as a scale of inertia for normalization or diagonal approximation, To approximate the components of the actuator output torque / force vector, For the first i The Coriolis force and the components of the centrifugal force term are approximately equal in degree of freedom. For the first i The components of the gravity term in the degrees of freedom are approximated. , and For component-based extended state observer gain, For the first i The derivative of the position estimation of degrees of freedom, For the first i The derivative of the velocity estimation for degrees of freedom, For the first i The derivative of the degree-of-freedom perturbation estimate, For the first i The actual joint positions of the degrees of freedom; Perturbation estimation in joint space dynamics model The specific expression is: In the formula, For the first i Degrees of freedom perturbation estimation , n For the number of degrees of freedom, It is the transpose symbol. .

4. The robot control method based on active perturbation observation and phase coupling control according to claim 3, characterized in that, In S3, the method of mapping the phase to the desired trajectory using phase parameterization is as follows: Introduce phase to the degrees of freedom and define the phase-to-desired trajectory mapping. The expression: In the formula, For offset position, The amplitude coefficient, Let be the basis functions from phase to position. For the first i Phase, specifying the phase rate The following coupled oscillator model is given: In the formula, No. i The natural frequency of degrees of freedom For the elements of the coupling matrix, by Regarding time t By integration, For phase adaptive gain, For the first j Phase, This is the phase error; In the formula, The desired phase is given for task planning; The calculation is performed using the adaptive coupling gain law, the specific expression of which is: In the formula, The time derivative of the coupling gain. and For learning rate, This is the stabilization coefficient. These are coupling prior values.

5. The robot control method based on active perturbation observation and phase coupling control according to claim 4, characterized in that, In S3, the specific expression for the closed-loop trajectory tracking and disturbance compensation control law is as follows: In the formula, These are the feedback gain matrix coefficients, used to drive the system state to converge to the sliding manifold. s The second sliding manifold is used to embed the position error transformation and its constraint characteristics into the control input. This is the amount of shock buffer injection specifically designed for the landing phase. Let the desired joint position vector be... Let the desired joint velocity vector be... Let be the desired joint acceleration vector.

6. The robot control method based on active perturbation observation and phase coupling control according to claim 5, characterized in that, Calculate the amount of impact buffer injection specifically for the landing phase using a landing impact buffering strategy. Its expression is as follows: In the formula, Estimate the relative velocity or contact velocity vector of the joint at the moment of contact. It is diagonally positive definite. is the scaling factor.

7. The robot control method based on active perturbation observation and phase coupling control according to claim 6, characterized in that, In S3, the method for calculating the second sliding manifold is as follows: Construct the first sliding manifold based on the position error transformation. Then, by combining auxiliary weights, a second sliding manifold is constructed. ; In the formula, This is the joint position tracking error vector. , This is the joint velocity tracking error vector. , This is the position-velocity coupling matrix. , For diagonalization operation, It is the first constant that makes up the coefficients of the matrix. The first of the coefficients of the matrix n constant; In the formula, For reference safe position vector, Diagonal auxiliary weight moments generated for task performance metrics.

8. The robot control method based on active perturbation observation and phase coupling control according to claim 7, characterized in that, In S4, the method for introducing position error transformation is as follows: Define the tracking error and calculate the position error transformation amount using the position constraint transformation function; In the formula, For the first i Free position error conversion amount, For the first i The maximum safety deviation allowed for degrees of freedom For the first i Position tracking error for degrees of freedom.

9. The robot control method based on active perturbation observation and phase coupling control according to claim 8, characterized in that, In S4, the specific method for designing a continuously differentiable robot-aided level function is as follows: Define a task performance function, and calculate the diagonal auxiliary weight moments using the task performance function. , , For the first i Task performance function with degrees of freedom; In the formula, To measure or estimate the force / torque components of human-computer interaction, and For the preset weights, based on Design a continuously differentiable robot-aided level function. Its expression is as follows: In the formula, The boundary of the dead zone This is the boundary of the saturation region. This is the slope coefficient of the hyperbolic tangent function, used to adjust the steepness of the function in the transition region. This is the center offset of the hyperbolic tangent function, used to determine the center position of the function's transition.

10. The robot control method based on active perturbation observation and phase coupling control according to claim 9, characterized in that, In S5, the Lyapunov method is specifically as follows: Define the tracking error and use the candidate Lyapunov function. The specific expression is: In the formula, To couple the target value or desired value, These are the coefficients of the position potential gain matrix in the Lyapunov function. For candidate Lyapunov functions Regarding the time derivative, substituting the closed-loop trajectory tracking and disturbance compensation control law and the coupling gain adaptive law into... The expression for , under the conditions of satisfying the dynamic assumptions and reasonable gain selection, can be obtained as follows: In the formula, Candidate Lyapunov functions The time derivative is used to evaluate the rate of energy change and stability of the system.

Images