Quadruped robot maneuvering control system and method with unknown interference adaptability

Abstract

The invention discloses a quadruped robot maneuvering control system and method with unknown interference adaptability, and belongs to the technical field of quadruped robots, and the method comprises the steps: building a quadruped robot dynamics model; a joint state disturbance observer is designed in combination with a quadruped robot kinetic model; the joint state interference observer is a nonlinear interference observer; integrating the joint state basic controller and the joint state interference observer to form a joint state anti-interference controller; and proving the Lyapunov stability of the joint state disturbance observer. According to the method, the feed-forward item is constructed by using the current information and added into the basic controller, the influence of unknown linear and nonlinear interference is actively compensated, and the influence of unknown interference on a dynamic system can be fundamentally solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of quadruped robots, and particularly relates to a quadruped robot mobile control system and method with unknown disturbance adaptability. BACKGROUND

[0002] Quadruped robots have shown great application potential in post-disaster rescue, field exploration, heavy load transportation and other fields due to their strong adaptability to complex terrains. However, the actual working environment of the quadruped robots is often full of multi-source disturbances, such as leg impact caused by protrusions and depressions in the field, dynamics parameter fluctuation caused by load changes, sudden changes in natural wind resistance and ground friction coefficient, and hardware inherent errors such as sensor noise and motor hysteresis. These disturbances directly affect the gait stability, motion accuracy and working safety of the quadruped robots. Therefore, constructing a control architecture with active disturbance suppression and real-time compensation capability is the core to improve environmental robustness. As the feedforward-feedback hub for realizing the architecture, the performance optimization of the disturbance observer has important scientific value and engineering significance.

[0003] The prior art has carried out a lot of explorations around unknown disturbance adaptation technology, forming various typical methods such as PID (proportional-integral-derivative control) control, MPC control and the like. The PID control parameters have strong dependence and weak dynamic adaptability. When the working environment of the quadruped robot changes greatly, the original parameters cannot match the new disturbance characteristics, and gait oscillation and response lag phenomena are likely to occur. The performance can be restored only by manually adjusting the parameters repeatedly, which is difficult to meet the dynamic operation demand, and there is an obvious defect in the adaptability to unknown disturbances. Therefore, the control method has strong inconvenience. Since the PID control method only corrects the error to make the output tend to the expected value, it does not actively compensate for the disturbance signal, and the disturbance is offset by the robustness of the system itself, which is difficult to compensate for the influence of the disturbance on the dynamic system from the root. Moreover, due to the highly nonlinear characteristics of the quadruped robot such as joint friction and motor hysteresis, the linear regulation logic of the PID cannot adapt to such nonlinear disturbances, and the control accuracy will be greatly reduced in high-speed motion or heavy load scenarios. The MPC (model predictive control) control is based on the system dynamics model to predict the system state in the future period of time, and outputs the optimal control amount by solving the constrained optimization problem. However, in the field of anti-disturbance, since the optimization effect of the MPC depends on the accurate dynamics model, when there is an unmodeled disturbance, the deviation between the model and the actual system will directly lead to the distortion of the prediction result, and then the control output deviates from the optimal solution. The MPC needs to solve a multivariable constrained optimization problem in each control period, and needs to optimize the joint torque of the four legs and the trunk posture at the same time, which has very high requirements for hardware computing resources. The disturbance often needs to be responded quickly, and the calculation delay of the MPC will cause the compensation to be not in time, missing the best opportunity to resist the disturbance. SUMMARY

[0004] To solve the above technical problems, the application adopts the following technical solutions:

[0005] A four-legged robot mobile control system with unknown interference adaptability, comprising: a joint state basic controller, an actuator, a four-legged robot platform, a joint sensor, a gyroscope and an accelerometer, a joint state disturbance observer, an expected joint angle generation module, and a communication unit; wherein the joint state basic controller, the joint state disturbance observer, and the expected joint angle generation module constitute a joint state anti-interference controller;

[0006] The four-legged robot platform is used to simulate a real four-legged robot in an experiment, and stores historical joint angle information and command information as input signals of the expected joint angle generation module;

[0007] The joint sensor is used to measure the measured joint angle information of the four-legged robot;

[0008] The gyroscope and the accelerometer are used to measure the attitude information of the four-legged robot;

[0009] The joint state disturbance observer is used to observe external disturbances and internal disturbances that cause the joints of the four-legged robot to deviate in angle; the output of the nonlinear joint state disturbance observer is added to the joint state basic controller in the form of a feedforward term;

[0010] The expected joint angle generation module takes the measured attitude information, the measured joint angle information, the historical joint angle information, and the command information of the four-legged robot as inputs, and takes the expected joint angle information as output;

[0011] The joint state basic controller obtains the joint motor control torque through the deviation information between the expected joint angle information and the measured joint angle information;

[0012] The communication unit sends the measured joint angle information to the joint state anti-interference controller, and converts the joint motor control torque into an instruction motor voltage signal before distributing it to the four-legged robot platform through the actuator.

[0013] A four-legged robot mobile control method with unknown interference adaptability, for the four-legged robot mobile control system with unknown interference adaptability, comprising:

[0014] Step 1, establishing a four-legged robot dynamics model;

[0015] Step 2, designing a joint state disturbance observer in combination with the four-legged robot dynamics model; the joint state disturbance observer is a nonlinear disturbance observer;

[0016] Step 3, integrating the joint state basic controller and the joint state disturbance observer to form a joint state anti-interference controller;

[0017] Step 4, joint state disturbance observer Lyapunov stability proof is carried out.

[0018] The present application has the following beneficial effects:

[0019] (1) The present application uses current information to construct a feedforward term to join the basic controller, actively compensating for the influence of unknown linear and nonlinear disturbances, and solving the influence of unknown disturbances on the dynamic system from the root, rather than relying on feedback information to passively suppress disturbances.

[0020] (2) The present application is a modular technology that embeds the deduction of the dynamics of a quadruped robot and the estimation technology of unknown disturbances, and can be independently embedded into the control framework of a quadruped robot to enhance the anti-interference ability of the system. Compared with existing anti-interference technologies, it can be more conveniently integrated into the controller and applied to actual needs.

[0021] (3) The present application has reduced dependence on the accurate model of the dynamics of a quadruped robot, and has reduced the complexity of the highly nonlinear dynamic modeling of a quadruped robot, reduced the occupation of computing resources, and has good unknown disturbance adaptation ability under the designed reasonable simplified dynamic model.

[0022] (4) The present application can quickly reflect the dynamic deviation, simplify the actual application adjustment parameter process, optimize many control algorithms, and improve the dynamic operation efficiency of a quadruped robot and the adaptability to unknown disturbances. For example, in a reinforcement learning control system, different reward functions and training frameworks are required to be designed according to the needs of different task scenarios to obtain the corresponding strategy for the current task scenario. After adding the disturbance observer, only the appropriate PID parameters need to be adjusted, and the disturbance observer parameters are independently adjusted to improve the generalization of the strategy, and the quadruped robot can still maintain stable and efficient motion under the influence of unknown disturbances that do not appear in the training environment, thereby enhancing the anti-interference ability of the quadruped robot in dynamic operation demand and the adaptability to unknown environments. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a flow chart of the anti-interference loop of the present application;

[0024] Figure 2 is a structural schematic diagram of the quadruped robot mobile control system with unknown disturbance adaptability of the present application, wherein 1 is a joint state basic controller, 2 is an actuator, 3 is a quadruped robot platform, 4 is a joint sensor, 5 is a gyroscope and accelerometer, 6 is a joint state disturbance observer, 7 is a desired joint angle generation module, and 8 is a communication unit. DETAILED DESCRIPTION

[0025] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not used to limit the present application. In addition, the technical features involved in the various embodiments of the present application described below can be combined with each other as long as they do not conflict with each other.

[0026] The present application provides a four-legged robot mobile control system and method with unknown disturbance adaptability, which can enhance the adaptability of the four-legged robot to unknown disturbances. The present application embeds a disturbance observer in the control system, which on the one hand enhances the anti-disturbance performance of the four-legged robot, and the observer can quickly and accurately estimate the disturbance suffered by the four-legged robot in the working environment, and through feedforward compensation, the anti-disturbance effect is achieved; on the other hand, through the embedding of the disturbance observer, the dynamic performance of the controller is optimized, and the time cost and labor cost of repeatedly adjusting a large number of parameters to adapt to the current working environment due to the change of the working environment in practical application are reduced. Taking reinforcement learning control as an example, when the disturbance observer is not embedded, the reward function needs to be designed for the current working environment, and once other tasks are involved, the adaptability of the original reward function decreases, which will lead to the decline of the motion performance of the four-legged robot, and the reward function needs to be redesigned and retrained for the new task scenario, which is time-consuming and labor-intensive. After embedding the disturbance observer, the strategy trained in the simple scene has strong generalization, and it can adapt to various unmodeled disturbances such as unknown terrain, additional load and training set speed command without repeated parameter adjustment, and it can work stably in multiple environments.

[0027] As Figure 1 shown is a flow chart of the anti-disturbance loop of the present application; the present application first establishes a four-legged robot dynamics model; then combines the joint state disturbance observer 6 designed by formula (12), (13), (14) to obtain the joint layer disturbance estimation value; finally, the above joint layer disturbance estimation value is introduced into the joint state basic controller 1 in the form of feedforward term, realizing online detection and compensation of compound disturbance.

[0028] Figure 2 The four-legged robot mobile control system with unknown disturbance adaptability of the present application comprises a joint state basic controller 1, an actuator 2, a four-legged robot platform 3, a joint sensor 4, a gyroscope and an accelerometer 5, a joint state disturbance observer 6, an expected joint angle generation module 7 and a communication unit 8; wherein the joint state basic controller 1, the joint state disturbance observer 6 and the expected joint angle generation module 7 constitute a joint state anti-disturbance controller.

[0029] The four-legged robot platform 3 is used to simulate the real four-legged robot in the experiment, and stores the historical joint angle information and command information as the input signal of the expected joint angle generation module 7.

[0030] The joint sensor 4 is used to measure the measured joint angle information of the quadruped robot. It uses a 14-bit absolute encoder to obtain the position and speed of each joint in real time.

[0031] The gyroscope and accelerometer 5, i.e., the IMU (Inertial Measurement Unit), are used to measure the attitude information of the quadruped robot as the input signal for the desired joint angle generation module 7.

[0032] The joint state disturbance observer 6 is used to observe external and internal disturbances that cause angular displacement of the quadruped robot joints; the output of the nonlinear joint state disturbance observer 6 is fed into the joint state basic controller 1 in the form of a feedforward term to realize online detection and compensation of joint disturbances.

[0033] The desired joint angle generation module 7 takes the measured posture information, measured joint angle information, historical joint angle information and command information of the quadruped robot as input, and the desired joint angle information as output.

[0034] The joint state basic controller 1 analyzes the deviation information between the expected joint angle information output by the expected joint angle generation module 7 and the current measured joint angle information output by the joint sensor 4, and processes the deviation information using a position control method (which can be proportional-derivative control, as shown in equation (15)) to obtain the joint motor control torque that causes the deviation value to tend to zero.

[0035] The actuator 2 consists of twelve joint motors of the quadruped robot. The joint motor control torque output by the joint state disturbance rejection controller is converted into command motor voltage signals, and the command motor voltage signals are distributed to the quadruped robot platform 3 through the actuator 2.

[0036] The communication unit 8 is used to realize communication between the quadruped robot platform 3 and the joint state disturbance rejection controller. It is used to send the measured joint angle information of the joint sensor 4 to the joint state disturbance rejection controller, and to feed back the command motor voltage signal of the joint state disturbance rejection controller to the quadruped robot platform 3 through the actuator 2.

[0037] The present invention provides a quadruped robot mobility control method with unknown disturbance adaptability, comprising:

[0038] Step 1: Establish a dynamic model of the quadruped robot;

[0039] The Lagrangian dynamics model of the joint layers of the quadruped robot is given:

[0040] (1)

[0041] in, These represent the joint angles, joint angular velocities, and joint angular acceleration vectors of the twelve joints of a quadruped robot, respectively. Represents the contact force Jacobian matrix; Indicates contact force; This refers to the generalized torque corresponding to the generalized joint state variables of the quadruped robot, i.e., the joint motor control torque. Indicates joint disturbance torque;

[0042] In equation (1), Represents the inertia matrix; Represents the matrix of inertial forces and Coriolis forces; Represents the gravity matrix; This represents the joint torque caused by contact forces; the specific expression is as follows:

[0043] Reflecting the system's inertial distribution, it is a symmetric positive definite matrix, expressed as:

[0044] (2)

[0045] in, They represent the first The mass, moment of inertia, translational Jacobian matrix, and rotational Jacobian matrix of each link; the superscript T throughout the text denotes the transpose matrix.

[0046] To improve the efficiency of the desired joint angle generation module 7, the following simplification strategy is adopted for equation (2): diagonalization assumption, ignoring the inertial coupling between joints, retaining only the diagonal term and the local joint inertia, reducing computational complexity; base inertia superposition, treating the base inertia as the translational and rotational inertia of the first three degrees of freedom. Using this simplification strategy, the diagonalization assumption ignores the coupling term, resulting in the dynamic coupling effect not being modeled, which may introduce estimation errors in high-dynamic motion, but the impact is acceptable for most simple motion forms; the simplified equation... as follows:

[0047] (3)

[0048] in, For the first Moment of inertia of each joint. These are the base's inertia in the x-direction, y-direction, and z-direction, which together constitute the base's rotational inertia. These are the rotational inertia of the left foreleg hip joint (lateral swing joint), the rotational inertia of the left foreleg hip joint (longitudinal swing joint), and the rotational inertia of the left foreleg knee joint.

[0049] Reflecting inertial and Coriolis forces, it consists of the time derivative of the inertial matrix and velocity coupling terms, its elements... Satisfying the equation:

[0050] (4)

[0051] in, This is the inertial coupling coefficient in robot dynamics, used to characterize the inertial coupling relationship between different joint movements. It is obtained by combining the partial derivatives of the inertial matrix elements. These represent the robot's inertia matrix. The Line number Column element, first Line number Column elements, , , All of these represent the joint indexes of the robot; For the first The generalized coordinates of each joint For the first The generalized coordinates of each joint For the first The generalized velocity of each joint, For the first The generalized velocity of each joint, Let be the dimension of the generalized variable space.

[0052] To improve the efficiency of the desired joint angle generation module 7 The following simplification strategy is adopted: the Coriolis force is approximated as linear damping, the complex velocity coupling square term is ignored, and only the linear damping term is retained, which is replaced by the damping coefficient B; this simplification strategy ignores the velocity square term and retains only the linear damping term, which is applicable to most low-speed motion scenarios, and the impact is acceptable (ignoring the velocity square term and retaining only the linear damping term is applicable to low-speed motion scenarios; this invention does not involve high-speed motion, and the impact is acceptable); simplified formula as follows:

[0053] (5)

[0054] , , , They are the same matrix, in diagonal form.

[0055] The gravity matrix, based on the position of the link's center of mass and the gravity projection, is expressed as:

[0056] (6)

[0057] in, For the first Jacobian matrix in the vertical direction of the centroid of each link. Indicates the link index of the robot; It is the acceleration due to gravity;

[0058] Let $\mathbf$ represent the torque of the contact forces at the ends of the feet of a quadruped robot projected onto the joint space through the Jacobian matrix at the ends of the feet. The projection torque of each foot contact force onto the joint space is calculated separately and then summed. The expression is:

[0059] (7)

[0060] in, Indicates the first Jacobian matrix at the foot of a leg; Indicates the first Contact force of the foot of the leg dimensional vector; These correspond to the left front leg, right front leg, left hind leg, and right hind leg, respectively. The expression is as follows:

[0061] (8)

[0062] in, Indicates the first The rotation axis of each joint; Indicates the first The vector from each joint to the foot. The value ranges from 1 to 12, corresponding to twelve joints; For from the first The force of the foot of one leg to the first Jacobian matrix of joints.

[0063] Step 2: Design a joint state disturbance observer 6 based on the quadruped robot dynamics model; the joint state disturbance observer 6 is a nonlinear disturbance observer.

[0064] Define an auxiliary variable :

[0065] (9)

[0066] in, For auxiliary functions, Indicates joint disturbance torque The estimated value of the interference;

[0067] Interference Observer The coefficient matrix satisfies the following conditions:

[0068] (10)

[0069] in, For time.

[0070] Combining equations (1), (9), and (10), we obtain:

[0071] (11)

[0072] In summary, the nonlinear joint state disturbance observer 6 is designed as follows:

[0073] (12)

[0074] in, For the gravity matrix, This refers to the generalized torque corresponding to the generalized joint state variables of the quadruped robot, i.e., the joint motor control torque.

[0075] Where auxiliary function Designed as follows:

[0076] (13)

[0077] in, The design constant determines the estimated update rate of the joint state disturbance observer 6; this is derived from equation (10). The expression:

[0078] (14)

[0079] in, It is the inverse of the inertia matrix M.

[0080] Output of nonlinear joint state disturbance observer 6 The joint state basic controller 1 is added in the form of a feedforward term to realize online detection and compensation of joint disturbances.

[0081] Therefore, a nonlinear joint state disturbance observer 6 is designed based on equations (12), (13), and (14).

[0082] Step 3: Integrate the joint state basic controller 1 and the joint state disturbance observer 6 to form a joint state disturbance rejection controller;

[0083] The joint state basic controller 1 and the joint state disturbance observer 6 are integrated. The disturbance estimate measured by the joint state disturbance observer 6 is added to the joint state basic controller 1 in the form of feedforward to form a joint state disturbance rejection controller.

[0084] The basic controller 1 for the joint states of the quadruped robot adopts proportional-derivative control:

[0085] (15)

[0086] in, The joint motor control torque is calculated based on the control law of the basic controller 1 before the addition of the disturbance observer. This is the proportionality coefficient. These are the differential coefficients. The desired joint angle information output by module 7 is the desired joint angle information.

[0087] Combining equations (12) and (15), the interference estimate is... By adding the feedforward term to the basic joint state controller 1, we obtain the joint state disturbance rejection controller:

[0088] (16)

[0089] Step 4: Perform Lyapunov stability verification for the joint state disturbance observer 6.

[0090] From equation (12), we can obtain:

[0091] (17)

[0092] in, This represents the difference between the actual interference and the estimated interference value. This represents the first derivative of the difference between the actual disturbance and the estimated disturbance with respect to time. It is the inverse of the inertia matrix M.

[0093] Constructing Lyapunov functions :

[0094] (18)

[0095] in, It is a unit array.

[0096] (19)

[0097] Depend on Characteristics are known It is a positive definite matrix, and >0;

[0098] (20)

[0099] Therefore, according to Lyapunov's stability theorem, the joint state disturbance observer 6 designed in equation (12) is globally asymptotically stable when hour, .

[0100] This invention discloses a quadruped robot motion control system and method with adaptability to unknown disturbances. Considering the quadruped robot's susceptibility to internal and external disturbances and model uncertainties, a quadruped robot dynamic model is established. Disturbance observers are designed at the joint layers of the quadruped robot to estimate disturbances caused by multiple sources. The control method, based on existing quadruped robot control methods, effectively suppresses and compensates for multi-source disturbances by embedding a joint state disturbance observer 6. A Lyapunov function is designed to prove its stability. This control method effectively improves the quadruped robot's anti-interference capability and motion performance, enhances its adaptability to unknown disturbances, and exhibits strong disturbance suppression capabilities in various task scenarios. This method runs in real-time on the joint state disturbance rejection controller of the quadruped robot platform 3, and, combined with existing control methods, achieves high-performance motion of the quadruped robot under conditions such as unknown terrain, additional loads, and high-speed commands. This invention, without altering the existing controller structure, achieves online detection and compensation of composite interference by embedding a joint state interference observer at the joint layer, ensuring system stability, improving the adaptability of quadruped robots to unknown interference during movement in different task scenarios, enhancing the controller's scalability, and improving the effectiveness and engineering practicality of quadruped robots in suppressing and compensating for composite interference. It can solve the problem of anti-interference control of quadruped robots in complex and unknown environments such as disaster relief, field exploration, and heavy-load transportation.

[0101] The above description is merely an embodiment of the present invention and does not limit the scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related system fields, are similarly included within the protection scope of the present invention.

[0102] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

Claims

1. A quadruped robot mobility control system with adaptability to unknown disturbances, characterized in that, include: The system includes a joint state basic controller, an actuator, a quadruped robot platform, joint sensors, a gyroscope and accelerometer, a joint state disturbance observer, a desired joint angle generation module, and a communication unit; among which, the joint state basic controller, the joint state disturbance observer, and the desired joint angle generation module constitute a joint state disturbance rejection controller. The quadruped robot platform is used to simulate real quadruped robots in experiments, storing historical joint angle information and command information as input signals for the desired joint angle generation module; Joint sensors are used to measure the measured joint angles of a quadruped robot; Gyroscopes and accelerometers are used to measure the attitude information of quadruped robots; The joint state disturbance observer is used to observe external and internal disturbances that cause angular displacement of the joints of a quadruped robot; the output of the nonlinear joint state disturbance observer is fed into the joint state basic controller in the form of a feedforward term. The desired joint angle generation module takes the measured posture information, measured joint angle information, historical joint angle information and command information of the quadruped robot as input, and the desired joint angle information as output. The joint state basic controller obtains the joint motor control torque by using the deviation information between the desired joint angle information and the measured joint angle information; The communication unit sends the measured joint angle information to the joint state disturbance rejection controller, and converts the joint motor control torque into a command motor voltage signal, which is then distributed to the quadruped robot platform through the actuator.

2. The quadruped robot mobility control system with unknown interference adaptability according to claim 1, characterized in that, The joint state basic controller analyzes the deviation between the expected joint angle information output by the expected joint angle generation module and the measured joint angle information output by the joint sensor, and processes the deviation information using a position control method to obtain the joint motor control torque that makes the deviation value tend to zero.

3. The quadruped robot mobility control system with unknown interference adaptability according to claim 1, characterized in that, The actuator consists of twelve joint motors of the quadruped robot. The joint state disturbance rejection controller outputs the joint motor control torque, which is converted into command motor voltage signals. These command motor voltage signals are then distributed to the quadruped robot platform through the actuator.

4. The quadruped robot mobility control system with unknown interference adaptability according to claim 1, characterized in that, The communication unit is used to enable communication between the quadruped robot platform and the joint state disturbance rejection controller. It is used to send the joint angle information measured by the joint sensor to the joint state disturbance rejection controller, and to feed back the command motor voltage signal of the joint state disturbance rejection controller to the quadruped robot platform through the drive mechanism.

5. A quadruped robot maneuver control method with unknown disturbance adaptability, used in a quadruped robot maneuver control system with unknown disturbance adaptability as described in any one of claims 1 to 4, characterized in that, include: Step 1: Establish a dynamic model of the quadruped robot; Step 2: Design a joint state disturbance observer based on the quadruped robot's dynamic model; The joint state disturbance observer is a nonlinear disturbance observer; Step 3: Integrate the joint state basic controller and the joint state disturbance observer to form a joint state disturbance rejection controller; Step 4: Prove the Lyapunov stability of the joint state disturbance observer.

6. The quadruped robot mobility control method with unknown disturbance adaptability according to claim 5, characterized in that, Step 1 includes: The Lagrangian dynamics model of the joint layers of the quadruped robot is given: （1） in, These are the joint angles, joint angular velocities, and joint angular acceleration vectors of the twelve joints of the quadruped robot; The contact force Jacobian matrix; Contact force; For joint motor control torque; For joint disturbance torque; The inertia matrix; The matrix represents the inertial force and Coriolis force. The gravity matrix; The joint torque is caused by the contact force; It is a symmetric positive definite matrix: （2） in, The first Mass, moment of inertia, translational Jacobian matrix, and rotational Jacobian matrix of each link; It consists of the time derivative of the inertia matrix and the velocity coupling term, and its elements Satisfying the equation: （4） in, This is the inertial coupling coefficient in robot dynamics, used to characterize the inertial coupling relationship between different joint movements. It is obtained by combining the partial derivatives of the inertial matrix elements. These represent the robot's inertia matrix. The Line 1 Column element, first Line 1 Column elements, , , All of these represent the joint indexes of the robot; For the first The generalized coordinates of each joint For the first The generalized coordinates of each joint For the first The generalized velocity of each joint, For the first The generalized velocity of each joint, The dimension of the generalized variable space; Let the moment of the contact force at the foot end of the quadruped robot be the torque projected onto the joint space through the Jacobian matrix at the foot end. Calculate the projected moment of each foot contact force in the joint space separately and then sum them. The expression is: （7） in, Indicates the first Jacobian matrix at the foot of a leg; Indicates the first Contact force of the foot of the leg dimensional vector; These correspond to the left front leg, right front leg, left hind leg, and right hind leg, respectively. for: （8） in, Indicates the first The rotation axis of each joint; Indicates the first The vector from each joint to the foot. The value ranges from 1 to 12, corresponding to twelve joints; For from the first The force of the foot of one leg to the first Jacobian matrix of joints.

7. The quadruped robot mobility control method with unknown interference adaptability according to claim 6, characterized in that, A simplification strategy is adopted for equation (2), resulting in a simplified equation. for: （3） in, For the first Moment of inertia of each joint. The moment of inertia of the base in the x-direction, the moment of inertia of the base in the y-direction, and the moment of inertia of the base in the z-direction together constitute the rotational moment of inertia of the base; These are the rotational inertia of the left foreleg hip joint (lateral swing joint), the rotational inertia of the left foreleg hip joint (longitudinal swing joint), and the rotational inertia of the left foreleg knee joint.

8. The quadruped robot mobility control method with unknown interference adaptability according to claim 6, characterized in that, Define an auxiliary variable : （9） in, For auxiliary functions, Indicates joint disturbance torque The estimated value of the interference; Interference Observer The coefficient matrix satisfies the following conditions: （10） in, For time; Combining equations (1), (9), and (10), we obtain: （11） The nonlinear joint state disturbance observer is designed as follows: （12） in, The gravity matrix; auxiliary functions for: （13） in, This is a design constant; derived from equation (10). : （14） in, It is the inverse of the inertia matrix M; The output of the nonlinear joint state disturbance observer By incorporating a joint state-based controller in the form of a feedforward term, online detection and compensation of joint disturbances can be achieved.

9. The quadruped robot mobility control method with unknown disturbance adaptability according to claim 8, characterized in that, Step 3 includes: The joint state basic controller and the joint state disturbance observer are integrated, and the disturbance estimate measured by the joint state disturbance observer is added to the joint state basic controller in the form of feedforward to form a joint state disturbance rejection controller. The basic controller for the joint states of the quadruped robot uses proportional-derivative control: （15） in, The joint motor control torque is calculated from the control law of the basic controller before the addition of the disturbance observer. This is the proportionality coefficient. These are the differential coefficients. The desired joint angle information is output by the module that generates the desired joint angle. Combining equations (12) and (15), the interference estimate is... By adding the feedforward term to the basic joint state controller 1, we obtain the joint state disturbance rejection controller: （16）。 10. The quadruped robot mobility control method with unknown interference adaptability according to claim 9, characterized in that, Step 4 includes: From equation (12), we get: （17） in, This represents the difference between the actual interference and the estimated interference value. This represents the first derivative of the difference between the actual disturbance and the estimated disturbance with respect to time. It is the inverse of the inertia matrix M; Constructing Lyapunov functions : （18） in, For unit array; （19） Depend on Characteristics are known It is a positive definite matrix, and >0; （20） The joint state disturbance observer designed according to equation (12) is globally asymptotically stable when hour, .

Images