CN-121973869-A - High dynamic motion control method and system for four-wheel foot robot

CN121973869ACN 121973869 ACN121973869 ACN 121973869ACN-121973869-A

Abstract

The invention discloses a high dynamic motion control method and a system for a four-wheel foot robot. The method comprises the steps of acquiring the state of a robot system through a sensor in real time, dynamically updating a linear prediction model and compensating nonlinear errors by adopting an adaptive time-varying model prediction control algorithm fused with affine compensation, solving expected foot end force on line, mapping the foot end force into hub driving moment and leg joint moment containing structural rigidization moment based on closed-chain jacobian mapping to cooperatively inhibit leg shearing deformation caused by high dynamic motion, and outputting calculated moment instructions to each executing motor. By the innovation of the algorithm level, the invention effectively solves the difficult problems of difficult consideration of the control precision, the system stability and the algorithm instantaneity of the wheel foot robot in a high dynamic scene on the premise of not changing hardware, and remarkably improves the motion performance of the robot under extreme working conditions such as rapid acceleration, heavy gradient climbing and the like.

Inventors

WU XINLIANG
ZHANG QINGYANG
XU HONGYI
YANG YUYAN
YU QINGYUN
SUN JUNHAN
YUAN YANG

Assignees

厦门理工学院

Dates

Publication Date: 20260505
Application Date: 20260202

Claims (10)

1. The high dynamic motion control method of the four-wheel foot robot is characterized by comprising the following steps of: A state sensing step of acquiring a robot system state estimation value in a current control period; A model prediction control step, based on the system state estimation value, calculating and obtaining an expected foot end force vector of the current period through a self-adaptive time-varying model prediction control algorithm fused with affine compensation; A moment collaborative mapping and compensating step, namely mapping the expected foot end force vector into a driving moment of a hub motor and a joint moment of a leg joint motor at the same time, wherein the joint moment comprises a structural rigidifying moment for resisting leg shearing deformation caused by wheel body driving force; A control output step of respectively outputting the driving moment and the joint moment to corresponding motor executors; And a circulation step, entering the next control period, and repeatedly executing the steps.
2. The method for controlling the high dynamic motion of the four-wheel-foot robot according to claim 1, wherein the adaptive time-varying model predictive control algorithm fused with affine compensation specifically performs the following procedures: The time-varying linearization process comprises the steps of taking a current system state estimated value and a control input at the last moment as reference working points, performing first-order Taylor expansion on a nonlinear dynamics equation of a robot to obtain a time-varying system matrix and an input matrix; The affine compensation term calculation process is used for calculating an affine compensation term for compensating linearization cutoff errors and nonlinear dynamics terms, wherein the affine compensation term is a difference value between a value of a nonlinear dynamics function at a current working point and an approximation value of a time-varying linearization model at the point; A prediction model construction process, namely constructing a discrete time domain prediction model containing the affine compensation term; and in the online solving process, a quadratic programming problem is constructed based on the prediction model, a preset cost function and constraint conditions, and the expected foot end force vector is output through online solving of a solver.
3. The method for controlling the high dynamic motion of the four-wheeled foot robot according to claim 1, wherein the process of mapping the desired foot end force vector into the joint moment is specifically that the desired foot end force vector is mapped based on the transpose of jacobian from the robot leg joint space to the foot end cartesian space, and the joint moment is obtained.
4. A method of high dynamic motion control for a four-wheeled foot robot according to claim 3, wherein the jacobian matrix is calculated based on the real-time configuration of the robot leg mechanism.
5. The method for controlling the high dynamic motion of the four-wheeled foot robot according to claim 1, wherein the process of mapping the expected foot end force vector to the hub driving moment is characterized by extracting a longitudinal force component in the expected foot end force vector and multiplying the longitudinal force component by the radius of the wheel to obtain the hub driving moment.
6. The method according to claim 1, wherein the model predictive control step and the moment cooperative mapping and compensation step are performed in a high frequency control cycle.
7. A method of high dynamic motion control for four-wheeled foot robots according to claim 3, characterized in that said method is adapted to a tandem legged foot robot by adjusting the kinematic parameters of the leg mechanism on which the jacobian matrix depends.
8. A high dynamic motion control system for a four-wheel foot robot for implementing the control method of any one of claims 1 to 7, the system comprising: the sensor module is used for acquiring the body gesture, angular velocity, joint angle and foot force information of the robot in real time; The state estimation module is connected with the sensor module and is used for processing the sensing information to generate a system state estimation value; The self-adaptive time-varying model prediction controller is connected with the state estimation module and is used for executing the algorithm process of claim 2 and outputting a desired foot end force vector; The power coupling mapping and compensating module is connected with the adaptive time-varying model prediction controller and is used for receiving the expected foot end force vector, calculating and outputting driving moment and joint moment based on jacobian matrix transposition mapping and longitudinal force component mapping; And the actuator driving module is connected with the power coupling mapping and compensating module and is used for sending a moment instruction to the hub motor and the leg joint motor.
9. A computer-readable storage medium having stored thereon a computer program, characterized in that the program, when executed by a processor, realizes the steps of the four-wheel foot robot high dynamic motion control method according to any one of claims 1 to 7.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the four-wheel foot robot high dynamic motion control method according to any one of claims 1 to 7 when the program is executed by the processor.

Description

High dynamic motion control method and system for four-wheel foot robot Technical Field The invention belongs to the technical field of robot motion control, and particularly relates to a high-dynamic motion control method and system for a four-wheel foot robot. Background The wheel foot robot is a novel mobile platform integrating wheel type mobile high-speed high-efficiency and leg type mobile strong obstacle crossing capability, and has great application potential in the fields of complex terrain inspection, emergency rescue, material delivery and the like. The core movement mode is to realize stable, rapid and strong-adaptability movement by cooperatively controlling the rotation of the wheel body and the movement of the leg joints. However, when the tasks such as rapid acceleration, rapid braking, high-speed turning, large-angle climbing, high-dynamic obstacle surmounting and the like are executed, the robot system presents strong nonlinearity and strong coupling characteristics, and extremely severe requirements are put on the stability, the precision and the instantaneity of the motion control method. At present, main stream control schemes in the field are mainly divided into two types, but all have obvious defects, and the requirements of the high dynamic motion scene are difficult to meet: The first type of scheme is wheel-leg function decoupling control, the scheme regards the wheel body and the leg of the robot as a subsystem with independent functions, and the wheel-ground contact force and the internal force of the leg structure are ignored by artificial function cleavage under the assumption that the wheel-hub motor only provides driving force in the advancing direction and the leg joint motor only provides supporting force in the vertical direction. This solution is applicable even at low speeds, but in high dynamic movements, the intense wheel-to-wheel interaction creates a large longitudinal shear moment. Because the leg controller is not given with an instruction for resisting the shearing moment, the leg mechanism can generate passive flexible deformation (such as unexpected folding or stretching) under the action of inertia, so that serious pitching or rolling oscillation of the machine body is caused, the posture is unstable, and accurate action cannot be completed. The second scheme is a control architecture based on model prediction, and a model prediction control method is introduced by researchers for improving control performance. The Linear Model Predictive Control (LMPC) generally performs discrete linearization on a nonlinear dynamic model at a certain stationary balance point (such as the level of the fuselage) of the robot (a continuous time model is subjected to discrete processing by adopting a first-order euler discretization method), and then solves a quadratic programming problem online. This approach is computationally efficient, but its linear model is only valid near the equilibrium point. When the robot performs actions such as large-angle climbing (such as a pitch angle exceeds 20 degrees), high-speed spin or falling recovery, the working state of the robot deviates from a preset linearization point, key parameters such as rotational inertia, foot moment arms and the like of the system are changed drastically, nonlinear factors such as Coriolis force, centrifugal force and the like become non-negligible, and a linear prediction model is severely distorted, so that control performance is reduced and even the system diverges. The other direction is to use Nonlinear Model Predictive Control (NMPC), which builds a predictive model based on complete nonlinear dynamics equations, and can theoretically describe the system characteristics accurately. However, NMPC requires online solving of large-scale nonlinear optimization problems, and the computational burden is extremely heavy. Limited calculation force of an onboard calculation unit (such as a mini industrial computer) of the wheel-foot robot is limited, the NMPC is difficult to achieve a 300 Hz-500 Hz high-frequency control period necessary for high dynamic movement, real-time requirements cannot be met, and practical engineering application is difficult to put into practical engineering application. In summary, the prior art has three dilemmas of instability, linear MPC model distortion and insufficient nonlinear MPC calculation force caused by decoupling control in a high dynamic scene, and the core contradiction is that the control precision, the system stability and the algorithm instantaneity cannot be considered. Therefore, an innovative control method is urgently needed, and high-precision, high-stability and high-frequency real-time control of a strong nonlinear and strong coupling dynamic system of the wheel-foot robot can be realized on an embedded platform with limited resources. Disclosure of Invention The invention aims to overcome the defects of the prior art and provides a high-dynamic motion control m