Hip exoskeleton control method with reaction force control and complementary constraints and medium

Abstract

The application provides a hip exoskeleton control method and system for reaction control and complementary constraint, and the method comprises the following steps: S1, constructing a hip exoskeleton man-machine system dynamics model, and establishing a coupled dynamics equation; S2, establishing a foot-ground contact nonlinear complementary constraint model, and introducing a relaxation parameter for further processing; S3, establishing a multi-objective reaction control cost function, comprehensively considering an exoskeleton space, a human space, an environment space and a complementary constraint violation penalty term, and constructing a quadratic cost function; and S4, solving a quadratic programming problem with a complementary constraint in real time, dynamically adjusting the relaxation parameter according to the real-time estimated stiffness of the ground and the gait phase, and outputting a control instruction. The application is based on the online estimation of the stiffness of the ground and the adaptive parameter adjustment mechanism of the gait phase, can adapt to different ground characteristics and gait stage changes in real time, dynamically adjusts the complementary relaxation parameter, and makes the system show stronger adaptability and robustness in a complex walking environment.

Description

Technical Field

[0001] This invention relates to the technical field of optimized control of lower limb exoskeleton robots, specifically to a hip joint exoskeleton control method and medium that combines reactive control and complementary constraints, and more particularly to a real-time collaborative control method and medium for a rigid hip joint exoskeleton that integrates reactive optimization and contact complementary physical modeling. Background Technology

[0002] Hip-jointed walking exoskeletons have broad application prospects in medical rehabilitation, military walking assistance, and industrial material handling. Current mainstream control methods mainly employ trajectory tracking control, impedance control, or a combination thereof. Trajectory tracking control is based on a predefined gait trajectory, requiring the exoskeleton joints to precisely track the reference angle; however, its rigid tracking characteristics easily lead to human-machine dynamic conflicts, resulting in unnatural movement and wearing discomfort. Impedance control establishes compliant human-machine interaction by simulating a spring-damping system; however, its fixed parameters make it difficult to adapt to different wearers and complex walking environments, and it is prone to sudden torque changes when switching foot-ground contact states.

[0003] Chinese invention patent document CN202511569248 discloses a hip exoskeleton control method based on gait pattern prediction and gait phase estimation. The method includes: constructing an initial unsupervised terrain detection model; training and testing the initial unsupervised terrain detection model using a dataset to obtain a target unsupervised terrain detection model; acquiring three-dimensional environmental point cloud data and preprocessing the three-dimensional environmental point cloud data to obtain a binary environmental image; inputting the binary environmental image into the target unsupervised terrain detection model, the target unsupervised terrain detection model outputting terrain detection results, and smoothing the terrain detection results to obtain a gait pattern prediction result; acquiring hip joint angular velocity, using the hip joint angular velocity as a teaching signal for an adaptive oscillator system to obtain an initial gait phase estimate, and correcting the initial gait phase estimate to obtain a target gait phase estimate; calculating a net assist torque based on the gait pattern prediction result and the target gait phase estimate, and switching the assist mode of the hip exoskeleton based on the net assist torque.

[0004] Regarding the aforementioned technologies, the inventors discovered that existing rigid hip-joint walking exoskeletons primarily rely on predefined gait pattern templates for control strategies, simplifying the foot-ground interaction process into discrete contact state switching. This method is prone to periodic fluctuations and momentary instability in control signals during dynamic walking, resulting in insufficient adaptability of the system to environmental disturbances. To address these shortcomings, there is an urgent need to construct a unified constraint model that deeply integrates the exoskeleton's motion planning capabilities with the continuous dynamic characteristics of foot-ground physical interaction. Based on this model, a control framework for real-time response and adaptive adjustment should be established to fundamentally improve the exoskeleton's motion accuracy and overall coordination in complex environments. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method and medium for controlling a hip exoskeleton using reactive control and complementary constraints.

[0006] A method for controlling a hip exoskeleton based on reactive control and complementary constraints according to the present invention includes: Step S1: Construct a dynamic model of the hip joint exoskeleton human-machine system and establish coupled dynamic equations; Step S2: Establish a nonlinear complementary constraint model for foot-ground contact and introduce relaxation parameters for further processing; Step S3: Based on the coupled dynamic equations, the nonlinear complementary constraint model of foot-ground contact, and the relaxation parameters, establish a multi-objective response control cost function, and construct a quadratic cost function by integrating the exoskeleton space, human space, environmental space, and the penalty term for violation of complementary constraints. Step S4: Based on the quadratic cost function and the nonlinear complementary constraint model of foot-ground contact, solve the quadratic programming problem with complementary constraints in real time, and dynamically adjust the relaxation parameters according to the real-time estimated stiffness of the ground and the real-time gait phase, and output control commands.

[0007] Preferably, the process of establishing the coupled dynamic equations includes: Define generalized coordinates

[0008] in The angle of the human hip joint. The hip joint drive angle of the exoskeleton; The system dynamics follow the Lagrange equations:

[0009] In the formula, Represents the mass inertia matrix. Represents the Coriolis force and centripetal force terms. Represents the gravity term. This represents the total torque.

[0010] Preferably, step S2 includes: Based on the vertical distance function from the foot to the ground, complementary constraints of normal and tangential contact are constructed, and relaxation parameters are introduced for constraint relaxation, discretization and linearization. Define the vertical distance function from the sole of the foot to the ground:

[0011] in n The ground normal vector, unit ground normal vector n transpose, For reference points on the ground; when When the soles of the feet are off the ground, It touches the ground.

[0012] Preferably, the strict form and relaxation of the complementary normal constraint include: Strict complementarity conditions are:

[0013] in This represents the normal component of the contact force; the meaning of this formula is: and At least one of them is zero; Its equivalent can be written as:

[0014] In numerical optimization, equality constraints This results in a non-convex feasible set and singular gradients, to which a relaxation parameter is introduced. Relax it into an inequality constraint: ; The establishment of tangential complementary constraints includes: Considering the Coulomb friction model, tangential force satisfy ,in The reaction force is the ground reaction force at the foot. The coefficient of friction is given; to handle the continuous transition from static to kinetic friction, a tangential velocity is introduced. And establish complementary relationships: ; Constraint discretization and linearization include: At sampling time Known , Constraints on decision variables at the current moment The place is satisfied; Normal constraint is written as:

[0015] Tangential constraints are written as:

[0016] in , .

[0017] Preferred quadratic cost function include: Exoskeleton Space Cost :

[0018] in, For trajectory tracking items, This refers to the actual angle of the exoskeleton joint. To track the desired angle; To control energy consumption, This provides the output torque for the motor. For impact suppression term, For joint acceleration, , , These are the weighting coefficients in the exoskeleton spatial cost function; Human space cost :

[0019] in, For human body load items, The active torque of the human body; For the naturalness of motion, The desired hip joint angular velocity; This is the weighting coefficient for the human body load item. This refers to the weighting coefficient for the naturalness of motion term; Environmental space cost :

[0020] in, For contact force tracking, For time-varying reference contact force; For stability, For the contact force tracking term, the weighting coefficient is... These are the weighting coefficients for the stability term; Zero torque point The calculation formula is:

[0021] in , , Let be the position coordinates of the i-th particle in the global coordinate system. , For the first i The external torque acting on a particle is x , y directional components, The acceleration due to gravity is constant. , The coordinates of the zero torque point in the horizontal plane; Complementary constraint violation penalty :

[0022] in These are tangential complementary relaxation parameters; This term incorporates complementary inequality constraints into the objective function as penalties. When the constraints are satisfied, the penalty is zero; when the constraints are violated, a corresponding quadratic penalty is introduced to drive the optimal solution back to the feasible region. These are the weighting coefficients; The total cost function is then: .

[0023] Preferably, step S4 includes solving a quadratic programming problem with complementary constraints in real time, discretizing the continuous-time optimization problem into a standard quadratic programming form at the sampling time.

[0024] Preferably, to achieve millisecond-level control, continuous-time optimization is discretized into linear or quadratic programming; at the sampling time... Known state and estimates The decision variables are ,but: Discretization of dynamic constraints includes: The dynamic equation at time Discretized and ignoring higher-order terms, the following linear equality constraints are obtained:

[0025] This formula is about Linear equality constraints; Discretization of other constraints includes: Motor output constraints: ; Contact force constraints: ; Friction cone constraint: ,in , ; Joint angle limit: predict the next position based on the current state. ,Require This is about Linear inequalities; Discretization of the cost function includes: At any moment , cost function Discretize each term in the equation, and after rearranging, the cost function is obtained. Represented as decision variables The quadratic form:

[0026] in It is a positive semi-definite Hessian matrix. This is the vector of linear term coefficients; The standard form of quadratic programming: Based on the above discretization steps, at time... The quadratic programming problem that needs to be solved in time is:

[0027] in A eq This is the equality constraint matrix; A ineq This is the inequality constraint matrix; By solving this quadratic programming problem, the optimal control input at the current moment can be obtained. .

[0028] Preferred complementary relaxation parameters The online adjustment includes the adjustment of the normal relaxation parameter and the adjustment of the tangential relaxation parameter.

[0029] Preferred normal relaxation parameters The adjustments include: Ground stiffness Estimated by the real-time relationship between contact force and foot displacement:

[0030] in This represents a small change; to avoid the denominator approaching zero during calculation, a filter is used for smoothing to ensure the stability of the estimated value; the adjustment law is designed as follows:

[0031] in, These are the initial normal relaxation parameters. For the estimated ground stiffness, To adjust the gain; Tangential relaxation parameters The adjustments include: Gait phase The pitch angle period signal from the thigh IMU is estimated and normalized to... ,in Corresponding to heel strike; the adjustment mechanism is designed as follows:

[0032] in, These are the initial tangential relaxation parameters. This represents the phase shift during the mid-cycle of the gait swing, approximately , To adjust the gain constant.

[0033] According to the present invention, a computer-readable storage medium storing a computer program is provided, wherein when the computer program is executed by a processor, the steps of the described reactive control and complementary constraint hip exoskeleton control method are implemented.

[0034] Compared with the prior art, the present invention has the following beneficial effects: 1. Improve walking flexibility and wearing comfort: This invention establishes a nonlinear complementary constraint model for foot-ground contact, describing the contact dynamics as a continuous and smooth constraint relationship. This avoids the torque abrupt changes and control signal fluctuations caused by the switching of traditional discrete contact states, significantly improving the naturalness of the walking process and the flexibility of human-computer interaction, and enhancing wearing comfort.

[0035] 2. Enhanced environmental adaptability and disturbance resistance: Based on the online estimation of ground stiffness and the adaptive parameter adjustment mechanism of gait phase, it can adapt to different ground characteristics (such as softness and hardness differences) and gait phase changes in real time, dynamically adjust complementary relaxation parameters, and enable the system to exhibit stronger adaptability and robustness in complex walking environments.

[0036] 3. Achieve multi-objective collaborative optimization control: By designing a unified multi-objective response control cost function, multiple objectives such as exoskeleton trajectory tracking, human body load reduction, contact force smoothing, and system stability are integrated and solved in real time under the quadratic programming framework, thus achieving collaborative optimization of exoskeleton motion performance, human body metabolic saving, and overall stability.

[0037] 4. Avoid the limitations of predefined gait templates: It gets rid of the dependence on fixed gait pattern templates and generates control commands in real time through reactive optimization. It can flexibly respond to human movement intentions and environmental changes, effectively avoiding control instability and human-machine conflict caused by template mismatch.

[0038] 5. Improve system real-time performance and practicality: Discretize the continuous-time optimization problem into an efficient quadratic programming form, which is suitable for millisecond-level real-time solutions, meets the real-time control requirements of exoskeleton systems, and has strong engineering practical value.

[0039] 6. Provide a general control framework for rigid exoskeletons: The proposed control method that integrates reactive optimization and contact complementary modeling is not only applicable to hip exoskeletons, but also provides a scalable theoretical basis and system framework for compliant control of other joints or whole-body exoskeletons. Attached Figure Description

[0040] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a flowchart of the hip exoskeleton control method based on reactive control and complementary constraints of the present invention.

[0041] Figure 2 This is a schematic diagram of the overall architecture of the hip exoskeleton control method based on reactive control and complementary constraints of the present invention.

[0042] Figure 3 This is a schematic diagram of complementary constraints for foot-to-ground contact.

[0043] Figure 4 This is a flowchart illustrating the implementation of closed-loop control based on reactive optimization and adaptive relaxation adjustment according to the present invention. Detailed Implementation

[0044] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0045] like Figure 4 As shown, a method for controlling a hip exoskeleton using reactive control and complementary constraints includes: Step S1: Construct a dynamic model of the hip joint exoskeleton human-machine system and establish coupled dynamic equations; Step S2: Establish a nonlinear complementary constraint model for foot-ground contact and introduce relaxation parameters for further processing; Step S3: Based on the coupled dynamic equations, the nonlinear complementary constraint model of foot-ground contact, and the relaxation parameters, establish a multi-objective response control cost function, and construct a quadratic cost function by integrating the exoskeleton space, human space, environmental space, and the penalty term for violation of complementary constraints. Step S4: Based on the quadratic cost function and the nonlinear complementary constraint model of foot-ground contact, solve the quadratic programming problem with complementary constraints in real time, and dynamically adjust the relaxation parameters according to the real-time estimated stiffness of the ground and the real-time gait phase, and output control commands.

[0046] like Figure 1 and Figure 2 As shown, in one specific embodiment, the specific steps include: A dynamic model of the hip exoskeleton human-machine system was constructed, and coupled dynamic equations involving the human lower limb and the hip exoskeleton were established. A nonlinear complementary constraint model for foot-to-ground contact is established. Based on the vertical distance function from the foot to the ground, complementary constraints for normal and tangential contact are constructed, and relaxation parameters are introduced for constraint relaxation, discretization, and linearization. Design a unified multi-objective response control cost function, integrate the exoskeleton space, human space, environmental space and complementary constraint violation penalty terms, and construct a quadratic cost function; Real-time solution of quadratic programming problems with complementary constraints, discretizing the continuous-time optimization problem into a standard quadratic programming form at sampling time; Based on environment and gait adaptive parameter adjustment, the complementary relaxation parameters of normal and tangential directions are adaptively adjusted online according to the real-time estimated stiffness of the ground and the real-time gait phase.

[0047] The construction of the hip joint exoskeleton human-machine system dynamic model, establishing coupled dynamic equations involving the human lower limb and the hip joint exoskeleton, specifically includes: Establish coupled dynamic equations involving the human lower limb and hip exoskeleton. Consider the system's motion in the sagittal plane and define generalized coordinates. ,in The angle of the human hip joint (relative to the pelvis). This refers to the hip joint drive angle of the exoskeleton. The system dynamics follow the Lagrange equation:

[0048] The specific derivations of each term in the formula are as follows: (1) Mass inertia matrix Derivation: The kinetic energy of the system is Consider the mass distribution of the human thigh, lower leg, foot, and exoskeleton driving links. A lumped mass model is used, with the center of mass of each link located... It can be calculated using forward kinematics:

[0049] in Let be the positive kinematic function of the i-th link.

[0050] The corresponding velocity Jacobian matrix is Let the mass of the i-th link be... The moment of inertia is Then the kinetic energy is:

[0051] in N The total number of links in the system. For the first The angle of each link For the first The angular velocity of each link, The angular velocity is Jacobi.

[0052] Therefore, the mass inertia matrix is ​​obtained:

[0053] For hip exoskeletons, the inertia of the thigh segment (including the connecting rod between the human thigh and the exoskeleton) is mainly considered, and is approximated as follows:

[0054] Among the diagonal terms , These are the principal inertia and off-diagonal terms, respectively. This indicates human-machine inertial coupling.

[0055] (2) Coriolis force and centripetal force terms Derivation: according to calculate

[0056]

[0057] in , , Mass inertia matrix The corresponding element, , , Generalized coordinate vector The , , One portion, Generalized velocity vector The Each component.

[0058] Vector form is For planar rotational motion, this term mainly manifests as the centrifugal force effect.

[0059] (3) Gravity term Derivation: System potential energy ,in Let be the gravitational acceleration vector (assuming the y-axis is vertically upward). The gravitational term represents the negative gradient of the potential energy.

[0060] (4) Friction term Modeling: Consider joint viscous friction and Coulomb friction:

[0061] in , These are the matrices of viscous friction coefficient and Coulomb friction coefficient, respectively. It is a symbolic function.

[0062] (5) Total torque Decomposition: The total torque consists of three parts:

[0063] in: Mapping matrix for output torque of exoskeleton motors It indicates that it only drives .

[0064] For the active hip joint torque of the human body, the mapping matrix .

[0065] The reaction force is the ground reaction force at the foot. Foot position The Jacobian matrix satisfies Its transpose maps the contact force into the joint space.

[0066] like Figure 3 As shown, a nonlinear complementary constraint model for foot-to-ground contact is established. Based on the vertical distance function from the foot to the ground, complementary constraints for normal and tangential contact are constructed, and relaxation parameters are introduced for constraint relaxation, discretization, and linearization processing, specifically including: The complementary relationship in contact stems from the physical fact that the contacting bodies are impermeable and can only transmit pressure. Define the vertical distance function from the sole of the foot to the ground:

[0067] in n The unit ground normal vector, unit ground normal vector n transpose, This serves as a reference point on the ground. When... When the soles of the feet are off the ground, It touches the ground.

[0068] (1) Strict form and relaxation of complementary normal constraints: Strict complementarity conditions are:

[0069] in This represents the normal component of the contact force. The meaning of this equation is: and At least one of them is zero.

[0070] Its equivalent can be written as:

[0071] In numerical optimization, equality constraints This results in the feasible set being non-convex and the gradient being singular.

[0072] Therefore, a relaxation parameter is introduced. Relax it into an inequality constraint:

[0073] This allows for the existence of tiny "penetration-force" products, extending the feasible region into a convex set, while guaranteeing that when... It approximates physical reality.

[0074] (2) Establishment of tangential complementary constraints: Considering the Coulomb friction model, tangential force Must meet ,in The reaction force is the ground reaction force at the foot. Let be the coefficient of friction. To handle the continuous transition from static to kinetic friction, a tangential velocity is introduced. And establish complementary relationships:

[0075] That is, when the contact point is relatively stationary ( When sliding occurs, the tangential force is provided by static friction; when sliding occurs ( The constraint allows for a finite scalar product of tangential force and tangential velocity, avoiding abrupt force changes. Relaxation parameters. Control the smoothness of the transition.

[0076] (3) Discretization and linearization of constraints: At sampling time Known , The constraints need to be applied to the decision variables at the current moment. The conditions are met.

[0077] Normal constraint is written as:

[0078] Tangential constraints are written as:

[0079] in , .

[0080] A unified multi-objective response control cost function is designed, integrating exoskeleton space, human space, environmental space, and complementary constraint violation penalties to construct a quadratic cost function, specifically including: Cost function The design of the three-space objective is based on a quadratic form to maintain the convexity of the problem.

[0081] The detailed components are as follows: (1) Spatial cost of exoskeleton :

[0082] Tracking item :in The desired joint angles are provided by the high-level gait generator. This item is used to drive the actual angle of the exoskeleton joint. Tracking Expectations To achieve accurate trajectory tracking.

[0083] Control energy consumption items This item minimizes the motor output torque. The square of the design aims to reduce the energy consumption of exoskeleton systems and optimize power utilization efficiency.

[0084] Impact suppression term This item measures joint acceleration. The square of the value is used as a penalty to filter out high-frequency impact components caused by ground collisions. This term is based on the derivation of the second-order system transfer function, where the torque... With acceleration By coupling the moments of inertia, the penalty term indirectly limits the rate of change of torque, thereby improving the smoothness of motion and reducing drastic torque fluctuations.

[0085] To adapt to different working environments or task requirements, the weighting coefficients in the exoskeleton spatial cost function ( , , This can be dynamically adjusted. The specific adjustment criteria and steps are as follows: The system first integrates high-level commands and multi-source sensor information to identify task modes, ground impact intensity, and terrain features in real time. Then, it obtains basic weight values ​​according to a predefined weight mapping table and dynamically adjusts the weights based on online monitoring of trajectory tracking errors and instantaneous impact events. The adjusted weights are normalized and then fed into a quadratic programming solver. Finally, the system periodically evaluates the adjustment effect based on indicators such as tracking accuracy, torque fluctuation, and wearing comfort, and uses online learning algorithms to iteratively optimize the weight mapping rules and sensitivity parameters, thereby continuously improving the exoskeleton's autonomous adaptability and comprehensive control performance in different task scenarios.

[0086] For example, in precision operation tasks, the position tracking error term can be improved. The weight is adjusted to ensure the joints are precisely aligned with the target position; however, in tasks involving heavy-duty walking or high impact, the weight can be appropriately increased. and The weighting emphasizes torque control and acceleration smoothness, thereby improving system stability and user comfort.

[0087] This dynamic weight adjustment mechanism can be updated in real time through user-defined strategies, enhancing the versatility and adaptability of the exoskeleton system in multiple tasks and scenarios.

[0088] (2) Human space cost :

[0089] in This is the weighting coefficient for the human body load item. This represents the weighting coefficient for the naturalness of motion term.

[0090] Human load items This is based on the active torque of the human body. The estimation is performed using the inverse dynamics method, and the calculation formula is as follows: ,in Here is the mass inertia matrix of the human lower limbs. The angular acceleration of the human hip joint. For the Coriolis force and centripetal force terms of the human lower limbs, The gravitational component of the lower limbs. This is a coupling term. Its purpose is to minimize the output torque required by human muscles. This reduces metabolic consumption and optimizes the synergistic effect between the exoskeleton system and the human body.

[0091] Naturalness of motion :in The desired hip joint angular velocity is generated by a real-time intent recognition system. This improves the physiological naturalness and comfort of gait by encouraging the body's joints to move in a natural velocity pattern, avoiding gait distortion caused by exoskeleton guidance.

[0092] (3) Environmental space cost :

[0093] in For the contact force tracking term, This represents the weighting coefficient for the stability term.

[0094] Contact force tracking item :in This is a time-varying reference contact force that can be generated based on the gait phase. Specifically, the reference contact force is small at the beginning of the support phase, gradually increases as the gait progresses, and then decreases again during the push-off phase. This design aims to smoothly track the difference between the actual contact force and the reference contact force, reducing ground impact and optimizing mechanical performance.

[0095] Stability Term : where the zero torque point The calculation formula is:

[0096] in , , Let be the position coordinates of the i-th particle in the global coordinate system. , For the first i The external torque acting on a particle is x , y directional components, The acceleration due to gravity is constant. , The coordinates of the zero torque point in the horizontal plane.

[0097] After simplification, in planar motion, ZMP Approximately the intersection of the total gravity line and the ground. Supporting polygon. Enclosed by the contact points of both feet. This item is calculated. ZMP To support polygon S closest distance to the boundary And maximize the negative square of the distance, thereby improving the system's stability and disturbance resistance under external disturbances.

[0098] (4) Penalty for violation of complementary constraints :

[0099] in These are the tangential complementary relaxation parameters.

[0100] This term adds a penalty to the objective function when the complementary inequality constraint is satisfied. The penalty is zero when the constraint is satisfied and zero when the constraint is violated (i.e., when the constraint is violated, the penalty is zero). or If the constraints are not satisfied, a corresponding quadratic penalty is introduced to drive the optimal solution back to the feasible region. To ensure that the constraints are approximately satisfied during the numerical solution process, weighting coefficients should be set. It is large enough to strengthen the inhibitory effect on behavior that violates the constraints.

[0101] In summary, the total cost function is:

[0102] Real-time solution of quadratic programming problems with complementary constraints, discretizing the continuous-time optimization problem into a standard quadratic programming form at sampling time, specifically including: To achieve millisecond-level control, continuous-time optimization needs to be discretized into linear or quadratic programming. At the sampling time... Known state and estimates The decision variables are , The exoskeleton motor output torque at the current sampling moment. The ground contact force at the current sampling moment, The generalized acceleration of the system at the current sampling moment.

[0103] (1) Discretization of dynamic constraints: The dynamic equation at time Discretized and ignoring higher-order terms, the following linear equality constraints are obtained:

[0104] This formula is about Linear equality constraints.

[0105] (2) Discretization of other constraints: Motor output constraints: (Linear inequalities).

[0106] Contact force constraints: (Component linear inequality).

[0107] Friction cone constraint: ,in , This is a second-order cone constraint, which can be approximated as a linear constraint: .

[0108] Joint angle limit: predict the next position based on the current state. ,Require This is about Linear inequalities.

[0109] Complementary constraints: such as the discrete form established in the nonlinear complementary constraint model of foot-ground contact.

[0110] (3) Discretization of the cost function: At any moment , cost function Discretize each item in the table, where The variables are known quantities or can be expressed as a linear combination of known quantities and decision variables. After simplification, the cost function... It can be represented as a decision variable. The quadratic form:

[0111] in It is a positive semi-definite Hessian matrix. This is the vector of coefficients for the linear terms.

[0112] (4) Standard form of quadratic programming (QP): Based on the above discretization steps, at time... The quadratic programming problem that needs to be solved in time is:

[0113] in A eq This is the equality constraint matrix, corresponding to the linear equality constraints that the system's dynamic equations must strictly satisfy. A ineq It is an inequality constraint matrix, corresponding to linear inequality constraints such as motor output limiting, non-negative contact force, friction cone, joint angle limiting, and complementary relaxation constraints.

[0114] By solving this quadratic programming problem, the optimal control input at the current moment can be obtained. .

[0115] Adaptive parameter adjustment based on environment and gait To improve system adaptability, complementary relaxation parameters are used. Online adjustments are required.

[0116] (1) Normal relaxation parameter Adjustment: Ground stiffness It can be estimated through the real-time relationship between contact force and foot displacement:

[0117] in This represents a small change. To prevent the denominator from approaching zero during calculation, a filter is used for smoothing to ensure the stability of the estimated value. The adjustment law is designed as follows:

[0118] in, These are the initial normal relaxation parameters. For the estimated ground stiffness, To adjust the gain.

[0119] The physical meaning is: on a hard surface ( On a larger scale, even a tiny penetration can generate a large reaction force, therefore the clearance needs to be appropriately relaxed. This is to avoid unnecessary oscillations in the optimizer when strict constraints are met.

[0120] (2) Tangential relaxation parameters Adjustment: Gait phase The pitch angle period signal from the thigh IMU can be estimated and normalized to... ,in Corresponding to heel strike. The adjustment mechanism is designed as follows:

[0121] in, These are the initial tangential relaxation parameters. This represents the phase shift during the mid-cycle of the gait swing, approximately , To adjust the gain constant, the above functional form ensures that during the mid-phase of the oscillation (i.e., the stage when the horizontal velocity of the foot tip is greatest),... Reaching its maximum value allows for a larger tangential force-velocity product during this phase, which is beneficial for dynamic adjustment and compliance control of the foot trajectory; while in the support phase (i.e. Close to 0 or )hour, Choose a smaller value to enhance static friction and ensure stable contact between the foot and the ground.

[0122] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the described reactive control and complementary constraint hip exoskeleton control method.

[0123] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, making the system and its various devices, modules, and units of this invention function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0124] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A method for controlling a hip exoskeleton using reactive control and complementary constraints, characterized in that, include: Step S1: Construct a dynamic model of the hip joint exoskeleton human-machine system and establish coupled dynamic equations; Step S2: Establish a nonlinear complementary constraint model for foot-ground contact and introduce relaxation parameters for further processing; Step S3: Based on the coupled dynamic equations, the nonlinear complementary constraint model of foot-ground contact, and the relaxation parameters, establish a multi-objective response control cost function, and construct a quadratic cost function by integrating the exoskeleton space, human space, environmental space, and the penalty term for violation of complementary constraints. Step S4: Based on the quadratic cost function and the nonlinear complementary constraint model of foot-ground contact, solve the quadratic programming problem with complementary constraints in real time, and dynamically adjust the relaxation parameters according to the real-time estimated stiffness of the ground and the real-time gait phase, and output control commands; Complementary relaxation parameters The online adjustment includes the adjustment of the normal relaxation parameter and the adjustment of the tangential relaxation parameter; Normal relaxation parameters The adjustments include: Ground stiffness Estimated by the real-time relationship between contact force and foot displacement: in This represents a small change; to avoid the denominator approaching zero during calculation, a filter is used for smoothing to ensure the stability of the estimated value; the adjustment law is designed as follows: in, These are the initial normal relaxation parameters. For the estimated ground stiffness, To adjust the gain; Tangential relaxation parameters The adjustments include: Gait phase The pitch angle period signal from the thigh IMU is estimated and normalized to... ,in Corresponding to heel strike; the adjustment mechanism is designed as follows: in, These are the initial tangential relaxation parameters. This represents the phase shift during the mid-cycle of the gait swing, and its value is... , To adjust the gain constant.

2. The hip exoskeleton control method with reactive control and complementary constraints according to claim 1, characterized in that, The process of establishing coupled dynamic equations includes: Define generalized coordinates in The angle of the human hip joint. The hip joint drive angle of the exoskeleton; The system dynamics follow the Lagrange equations: In the formula, Represents the mass inertia matrix. Represents the Coriolis force and centripetal force terms. Represents the gravity term. This represents the total torque.

3. The hip exoskeleton control method with reactive control and complementary constraints according to claim 1, characterized in that, Step S2 includes: Based on the vertical distance function from the foot to the ground, complementary constraints of normal and tangential contact are constructed, and relaxation parameters are introduced for constraint relaxation, discretization and linearization. Define the vertical distance function from the sole of the foot to the ground: in n The ground normal vector, unit ground normal vector n transpose, As a reference point on the ground, P c Indicates the location of the sole of the foot; when When the soles of the feet are off the ground, It touches the ground.

4. The hip exoskeleton control method with reactive control and complementary constraints according to claim 3, characterized in that, The strict form and relaxation of complementary normal constraints include: Strict complementarity conditions are: in This represents the normal component of the contact force; the meaning of this formula is: and At least one of them is zero; Its equivalent can be written as: In numerical optimization, equality constraints This results in a non-convex feasible set and singular gradients, to which a relaxation parameter is introduced. Relax it into an inequality constraint: ； The establishment of tangential complementary constraints includes: Considering the Coulomb friction model, tangential force satisfy ,in The reaction force is the ground reaction force at the foot. The coefficient of friction is given; to handle the continuous transition from static to kinetic friction, a tangential velocity is introduced. And establish complementary relationships: ； Constraint discretization and linearization include: At sampling time Known , Constraints on decision variables at the current moment The place is satisfied; Normal constraint is written as: Tangential constraints are written as: in , .

5. The hip exoskeleton control method with reactive control and complementary constraints according to claim 1, characterized in that, Quadratic cost function include: Exoskeleton Space Cost : in, For trajectory tracking items, This refers to the actual angle of the exoskeleton joint. To track the desired angle; To control energy consumption, This provides the output torque for the motor. For impact suppression term, For joint acceleration, , , These are the weighting coefficients in the exoskeleton spatial cost function; Human space cost : in, For human body load items, The active torque of the human body; For the naturalness of motion, The desired hip joint angular velocity; This is the weighting coefficient for the human body load item. This refers to the weighting coefficient for the naturalness of motion term; Environmental space cost : in, For contact force tracking, For time-varying reference contact force; For stability, For the contact force tracking term, These are the weighting coefficients for the stability term; Zero torque point The calculation formula is: in , , Let be the position coordinates of the i-th particle in the global coordinate system. , For the first i The external torque acting on a particle is x , y directional components, The acceleration due to gravity is constant. , The coordinates of the zero torque point in the horizontal plane; Complementary constraint violation penalty : in These are tangential complementary relaxation parameters; This term incorporates complementary inequality constraints into the objective function as penalties. When the constraints are satisfied, the penalty is zero; when the constraints are violated, a corresponding quadratic penalty is introduced to drive the optimal solution back to the feasible region. These are the weighting coefficients; The total cost function is then: 。 6. The hip exoskeleton control method with reactive control and complementary constraints according to claim 1, characterized in that, Step S4 includes solving the quadratic programming problem with complementary constraints in real time, discretizing the continuous-time optimization problem into a standard quadratic programming form at the sampling time.

7. The hip exoskeleton control method with reactive control and complementary constraints according to claim 6, characterized in that, To achieve millisecond-level control, continuous-time optimization is discretized into linear or quadratic programming; at the sampling time... Known state and estimates The decision variables are ,but: Discretization of dynamic constraints includes: The dynamic equation at time Discretized and ignoring higher-order terms, the following linear equality constraints are obtained: This formula is about Linear equality constraints; Discretization of other constraints includes: Motor output constraints: ; Contact force constraints: ; Friction cone constraint: ,in , ; Joint angle limit: predict the next position based on the current state. ,Require This is about Linear inequalities; Discretization of the cost function includes: At any moment , cost function Discretize each term in the equation, and after rearranging, the cost function is obtained. Represented as decision variables The quadratic form: in It is a positive semi-definite Hessian matrix. This is the vector of linear term coefficients; The standard form of quadratic programming: Based on the above discretization steps, at time... The quadratic programming problem that needs to be solved in time is: in A eq This is the equality constraint matrix; A ineq This is the inequality constraint matrix; By solving this quadratic programming problem, the optimal control input at the current moment can be obtained. .

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the hip exoskeleton control method of reactive control and complementary constraints as described in any one of claims 1 to 7.

Images