Safety constraint optimization gait planning method and device for lower limb rehabilitation robot

Abstract

The invention discloses a safety constraint optimization gait planning method and device for a lower limb exoskeleton rehabilitation robot, and the method comprises the steps: designing an ACPG (adaptive central pattern generator) network topology for sagittal plane activity space limitation, and achieving the generation of a personalized gait track with a biological rhythm under human-computer interaction; on the basis of interaction torque and joint speed fed back by sensing, energy of human body active movement injected into the exoskeleton is calculated; constructing an intermediate variable according to the energy weighting, designing a gait adjustment model in combination with an admittance model, and outputting a self-adaptive smooth gait frequency and an amplitude modulation factor; based on the gait adjustment model, by using a control barrier function optimization theory, gait frequency and amplitude change are constrained within a safety range, so that the ACPG outputs rhythm gaits and meets safety requirements at the same time. According to the method, the energy change under the human-computer interaction effect is considered, and the personalized gait track with safety protection is generated.

Description

Technical Field

[0001] This invention belongs to the field of gait generation technology for lower limb exoskeleton robots, specifically relating to a safety constraint optimization gait planning method and device for a lower limb rehabilitation robot. Background Technology

[0002] Lower limb rehabilitation exoskeleton robots, by providing precise motor assistance, have become a key technology for neurological function reconstruction in patients with hemiplegia and spinal cord injury after stroke. Statistics show that millions of people worldwide suffer from walking dysfunction due to neurological diseases, while traditional rehabilitation training is limited by the physical strength and experience of therapists, making it difficult to achieve high-intensity personalized training. Exoskeleton rehabilitation robots are repeatable and controllable, but the interaction and matching between the exoskeleton robot and the wearer is difficult to adjust effectively. The human-machine gait adaptation capability of the exoskeleton directly determines the rehabilitation effect.

[0003] Current gait planning methods mainly include pre-programmed trajectory tracking, sensor-feedback-based impedance control, and bio-inspired Central Pattern Generator (CPG) models. Among these, CPG models simulate the rhythm-generating network in the biological central nervous system and are widely used in gait control and movement pattern generation. However, CPG models typically require complex parameter tuning and calibration to adapt to the gait characteristics of different users, which increases the complexity of their use.

[0004] Despite progress, existing methods still have the following limitations: First, traditional CPG parameter tuning mechanisms rely on offline optimization, making it difficult to respond in real time to changes in the wearer's active torque, resulting in asynchronous human-machine movement; second, most safety constraint strategies use fixed thresholds, ignoring individual differences in joint range of motion; more importantly, current gait planning schemes are mostly designed for structured environments, and their rigid boundary conditions may inhibit the flexible interaction characteristics required for rehabilitation training. These shortcomings may lead to joint overload risks and even secondary injuries.

[0005] Therefore, how to achieve personalized gait generation that ensures wearers can move within a safe space is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] To overcome the aforementioned shortcomings of existing technologies and adapt to the safety constraint requirements of gait generation in lower limb exoskeleton rehabilitation robots, this invention provides a safety constraint-optimized gait planning method and apparatus for lower limb rehabilitation robots. This invention not only improves the adaptability of the gait generation strategy but also ensures that the gait trajectory generated by the system is within the safety constraint space.

[0007] To address the aforementioned technical challenges, the first aspect of this invention relates to a gait adaptive planning method for a lower limb rehabilitation robot, comprising the following steps:

[0008] S1. Collect the gait trajectory of the hip and knee joints of one side of the human lower limb in the sagittal plane, and fit the trajectory at the hip and knee joints into mathematical expressions composed of two sine functions to obtain normal gait trajectory parameters; design an adaptive central pattern generator (ACPG) network topology structure constructed by four oscillators connected in series to adapt to the biomimetic structure of the human lower limb, and input the normal gait trajectory parameters into the ACPG model to generate the initial gait trajectory control lower limb exoskeleton rehabilitation robot;

[0009] The ACPG network has four oscillators arranged in a series chain. The rhythmic signals generated by the spontaneous oscillation of the first and second oscillators represent the hip joint control trajectory, while the rhythmic signals generated by the spontaneous oscillation of the third and fourth oscillators represent the knee joint control trajectory.

[0010] Four oscillators spontaneously oscillate based on the input gait trajectory parameters to generate rhythmic signals. Adjacent oscillators are coupled and coordinated to achieve smooth gait transitions.

[0011] S2. Real-time acquisition of interaction torque and joint angular velocity information generated by the wearable lower limb exoskeleton rehabilitation robot, and calculation of the energy injected into the exoskeleton robot by the active movement of the human lower limb; S3. Weighted combination of the energy generated by the interaction torque calculated in step S2 at the joint of one lower limb, and use the weighted energy value as the input value of the admittance model, and feed the output value of the admittance model back to the ACPG model as the frequency and amplitude factors of the gait trajectory, and update the gait trajectory online in real time;

[0012] The admittance model compliantly adjusts the frequency and amplitude factors of the feedback to the ACPG model, forming a personalized dynamic feedback.

[0013] S4. Construct a safety constraint set for the frequency and amplitude factors of the new gait trajectory. Based on the admittance model in step S3, and combined with the safety filter characteristics of the control barrier function CBF, constrain the frequency and amplitude changes of the gait within the specified range. The safety constraint set is explicitly expressed. Construct a quadratic programming function to solve for the new frequency and amplitude factors, and feed it back into the ACPG model to generate a personalized gait trajectory with safety assurance.

[0014] Specifically, step S1 includes:

[0015] S11 restricts the range of motion of the lower limb exoskeleton to the sagittal plane, making the hip and knee joints active joints for flexion and extension movements, and the ankle joint designed as a passive joint. It collects and fits the gait trajectory of the hip and knee joints of one side of the human lower limb in the sagittal plane.

[0016] S12 constructs the ACPG network topology, which consists of four closely coupled oscillators linked in a series chain. Each oscillator can oscillate spontaneously, generating a sinusoidal signal characterizing the motion trajectory of the corresponding joint; the motion trajectories of the hip and knee joints are respectively formed by the superposition of the sinusoidal signals output by the two corresponding oscillators.

[0017] S13 The first and second nodes jointly generate rhythmic signals that drive hip joint movement through internal coupling and synchronization mechanisms; the third and fourth nodes output in coordination with the same mechanism to form the knee joint movement trajectory; the four nodes cooperate with each other in the serial topology through phase coordination and amplitude modulation to finally synthesize a personalized gait trajectory that conforms to the characteristics of human movement.

[0018] Furthermore, step S12 specifically includes:

[0019] S121. Design a network topology consisting of 4 oscillators, with adjacent oscillators connected by arrows to form a coupling relationship;

[0020] S122. Each oscillator corresponds to a fundamental frequency term in a Fourier series;

[0021] S123. The target trajectory of the hip joint is formed by the superposition of the outputs of the first oscillator and the second oscillator, and the target trajectory of the knee joint is formed by the superposition of the outputs of the third oscillator and the fourth oscillator;

[0022] Step S13 specifically includes:

[0023] S131. Based on Fourier series theory, the hip / knee joint motion curve is approximated using the preset reference trajectory defined by equation (1):

[0024] (1)

[0025] S132. Establish the ACPG dynamic model as shown in equation (2):

[0026] (2)

[0027] in It is the first or A sine function, variable , , They represent the first The phase, amplitude, and offset of each oscillator; , , Represent the first term in equation (1) The frequency, amplitude, and offset of the state variables of a sin function; Determines the coupling strength; Determines the convergence speed; Represents scaled phase difference; discrete set Indicates the first A set of nodes from which a node receives information from its neighboring nodes; This represents the output of each node, and the trajectory of each joint can be obtained by superimposing the outputs of the nodes. The evolution of the phase over time is represented by the first equation in equation (2).

[0028] S133. Joint trajectory output satisfies equation (3):

[0029] (3)

[0030] S134. By adjusting factors and Adjust the frequency and amplitude of the gait. Therefore, adjustment factors are added to the second and fourth formulas of equation (2), respectively. and Perform gait trajectory planning.

[0031] Specifically, step S2 includes:

[0032] S21. Real-time monitoring of joint interaction torques in human-machine coupled motion. With joint angular velocity ;

[0033] S22. Calculate the mechanical power of the human lower limb active injection exoskeleton system based on the function-energy conversion principle, and calculate the energy transfer value of a single joint according to formula (4):

[0034] (4)

[0035] S23. By combining multi-joint energy weights, the wearer's movement intention is captured based on the calculated energy value generated by human-computer coupling interaction.

[0036] Specifically, step S3 includes:

[0037] S31 Constructs intermediate variables based on human-computer interaction energy;

[0038] S32 Establish the mass-spring-damper type admittance transfer function;

[0039] S33 outputs the frequency factor and amplitude factor for adaptive gait adjustment.

[0040] Specifically, step S31 includes:

[0041] S311. Use energy-weighted combination and Construct intermediate variables, and It is adjusted as a weighting factor for the torque of each joint during the entire gait pattern generation process;

[0042] S312. Define frequency-dependent energy input , It is the energy input value of the frequency admittance model;

[0043] S313. Define amplitude-dependent energy input , It is the energy input value of the amplitude admittance model.

[0044] Step S32 specifically includes:

[0045] S321. Establish the frequency admittance model:

[0046] (5)

[0047] S322. Establish the amplitude admittance model:

[0048] (6)

[0049] in, and It is the virtual inertia coefficient. and It is the virtual damping coefficient. and It is a virtual stiffness coefficient, and the admittance parameter can be adjusted according to different application scenarios and objectives.

[0050] The implementation of step S33 includes: through human-computer interaction energy Frequency factors affecting motion trajectory and amplitude factor Based on this characteristic, the wearer utilizes all the features of the exoskeleton. The gait trajectory of each joint is achieved by applying acceleration or deceleration torque to physically demonstrate their compliance or resistance.

[0051] Specifically, step S4 includes:

[0052] S41. Define the gait frequency modulation factor and amplitude modulation factor The set of safety constraints;

[0053] S42. Construct forward invariant constraints based on the control barrier function (CBF);

[0054] S43. Solve for the safe energy input value using quadratic programming.

[0055] Furthermore, step S41 specifically includes:

[0056] S411. Establish the constraint set Limited frequency factor and amplitude factor Feasible range:

[0057] (7)

[0058] S412 Construct constraint functions (8) and (9):

[0059] (8)

[0060] (9)

[0061] S413. Transform the constraint set into function form:

[0062] (10)

[0063] Step S42 specifically includes:

[0064] S421. Derive new constraints through relative degree analysis:

[0065] (11)

[0066] (12)

[0067] (13)

[0068] S422. Design forward invariance conditions (14)-(15):

[0069] (14)

[0070] (15)

[0071] S423. Calculate the relevant derivative terms (16)-(19):

[0072] (16)

[0073] (17)

[0074] (18)

[0075] (19)

[0076] Step S43 specifically includes:

[0077] Constructing quadratic programming problems (20)-(22):

[0078] (20)

[0079] (twenty one)

[0080] (20)

[0081] in, This represents the energy value calculated based on estimated torque and joint velocity. This means that the energy value that meets the safety constraints is obtained through quadratic programming, which can guarantee the frequency factor. and amplitude factor Within safe and feasible limits.

[0082] The method of this invention may further include step 5: To verify the safety constraint optimization gait planning method for lower limb rehabilitation robots designed based on the above steps, an ACPG network model and a lower limb exoskeleton dynamic model are built in Matlab2022a / Simulink, and simulation conditions are constructed to test the proposed method. Specifically, this includes:

[0083] S51. Construct patient muscle strength change functions: hip joint (23), knee joint (24):

[0084] (twenty three)

[0085] (twenty four)

[0086] S52 ACPG model parameters: Table 1 and Equation (25):

[0087] Table 1 ACPG Model Parameters

[0088] node 1 -0.226 0.066 -0.29 2 -0.083 0.066 -0.64 3 0.413 -0.225 1.02 4 -0.240 -0.225 0.41

[0089] (25)

[0090] The parameters of the admittance model related to the amplitude variation factor in S53 are set as follows: , , The admittance model parameters related to the frequency variation factor are set as follows: , , .

[0091] The parameters for the S54 lower limb exoskeleton dynamic model are set as follows: , , , , , , .

[0092] S55 Meanwhile, using CBF to adjust the deformation factor and When applying constraints, it is necessary to select an appropriate CBF function. In this experiment, all CBF functions selected are linear functions, i.e. , , , .

[0093] The frequency and amplitude constraints in S56 ACPG are respectively , , , .also, and These are the energy inputs for the two admittance models, and the weighted sums of the energies of the two joints, with the weights being: , , , .

[0094] The S57 PID controller serves as the underlying position controller for the lower limb exoskeleton robot. The PID parameters at the hip joint are as follows: , , The PID parameters at the knee joint are: , , .

[0095] A second aspect of the present invention relates to a gait safety planning device for a lower limb rehabilitation robot, comprising a memory and one or more processors, wherein the memory stores executable code, and the one or more processors execute the executable code to implement a gait safety planning method for a lower limb rehabilitation robot according to the present invention.

[0096] A third aspect of the invention relates to a computer-readable storage medium having a program stored thereon that, when executed by a processor, implements a gait safety planning method for a lower limb rehabilitation robot according to the invention.

[0097] This invention addresses the personalization and safety issues of gait planning for rehabilitation robots, proposing a gait safety planning method based on ACPG networks and CBF theory. This method generates personalized gait trajectories with safety protection by considering energy changes under the influence of human-computer interaction.

[0098] The innovation of this invention is:

[0099] 1. Propose ACPG-CBF fusion architecture: Propose a collaborative optimization framework based on ACPG network and CBF, which for the first time dynamically couples biological rhythm gait generation with real-time safety constraints, and solves the conflict between personalized adaptation and safety in traditional methods.

[0100] 2. Human-computer interaction energy-driven mechanism: Based on the "interaction torque-joint velocity" energy quantification model, the mechanical power injected into the exoskeleton by the active human movement is calculated, and the wearer's movement intention is captured in real time. This replaces the traditional fixed threshold safety strategy and breaks through the limitations of individual joint range of motion.

[0101] 3. Admittance-CBF dual-level safety regulation: A gait modulation factor is generated based on an energy-weighted admittance model. Combined with CBF quadratic programming, a dynamic safety barrier is constructed to strictly constrain the gait frequency / amplitude within the physiologically feasible domain, thereby achieving flexible interaction under rigid boundaries.

[0102] The working principle of this invention is:

[0103] 1. Biorhythm Adaptive Generation: The ACPG network topology uses Fourier series decomposition of the gait periodic function and dynamically fits the hip / knee joint trajectory using an oscillator coupling model. Its core parameters are controlled in real time by the output of the admittance model, giving the gait biorhythmic characteristics that can be adjusted online.

[0104] 2. Energy-Gait Mapping Mechanism: High-precision torque sensors collect joint interaction torques and velocities, and the energy input of each joint is calculated based on the function-energy conversion principle. The weighted combination of energy values ​​drives the admittance model, which transforms the wearer's intention into gait tuning commands, realizing a closed-loop mapping of "human active force → gait parameters".

[0105] 3. CBF Safety Constraint Implementation: A constraint set of frequency / amplitude factors is established, and the state-space constraints are transformed into differential inequalities by constructing a control barrier function. Quadratic programming is used to correct the admittance input energy in real time, forcing evolution within a safe range and eliminating the risk of joint overload from a dynamic perspective.

[0106] The beneficial effects of this invention are reflected in:

[0107] (1) Unlike the general practice of using a preset gait trajectory for rehabilitation training, this invention uses an ACPG network to generate a gait trajectory with rhythmic characteristics. The periodic changes of the gait trajectory are described by a mathematical model, which facilitates the real-time adjustment of the frequency and amplitude of the gait trajectory.

[0108] (2) Considering the human-computer interaction mechanism when the wearer uses the rehabilitation robot for gait training, calculate the energy generated by the human active interaction torque doing work on the lower limb exoskeleton robot, and design an adaptive adjustment model for gait frequency and amplitude in combination with the admittance model.

[0109] (3) While realizing adaptive gait adjustment, this invention considers the safety constraint of gait. Based on CBF theory, it constrains the frequency and amplitude changes of gait within a specified range and ensures the safety of the wearer under gait changes according to the wearer's movement state.

[0110] In summary, the gait planning method proposed in this invention considers both the personalized generation problem under human-computer interaction and the safety problem of gait generation, generating a rehabilitation gait that meets the wearer's needs. Attached Figure Description

[0111] To more clearly illustrate the technical solution of this invention, the appendices mentioned in the technical description of this invention are described below. Figure 1-10 A brief description is provided. Obviously, the accompanying drawings described below are only schematic diagrams of some embodiments of the present invention. Those skilled in the art can derive other similar illustrations from these drawings without any inventive effort.

[0112] Figure 1 This is a topology diagram of the ACPG model designed based on the structure of the lower limb exoskeleton rehabilitation robot of the present invention, wherein network nodes numbered 1-4 represent ACPG oscillator units;

[0113] Figure 2 This is a torque diagram of the human lower limbs acting on the hip and knee joints of the rehabilitation robot in the simulation, as described in this invention.

[0114] Figure 3 This is a diagram showing the energy changes of different joints calculated in simulation based on the active torque and velocity at the joints, as described in this invention.

[0115] Figure 4 The energy value input to the admittance model after weighted calculation. and The change graph;

[0116] Figure 5 Gait amplitude adjustment factor in ACPG models subject to specified range constraints The change graph;

[0117] Figure 6 Gait frequency adjustment factor in the ACPG model subject to specified range constraints The change graph;

[0118] Figure 7The images show the hip joint trajectory generated with and without CBF constraints.

[0119] Figure 8 The images show the knee joint trajectory generated after CBF constraint and without constraint optimization.

[0120] Figure 9 Gait trajectory tracking diagrams for the hip and knee joints of the lower limb exoskeleton;

[0121] Figure 10 This is a flowchart of the method of the present invention. Detailed Implementation

[0122] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0123] Example 1

[0124] This embodiment relates to a gait safety planning method applied to a lower limb rehabilitation robot, such as... Figure 10 The specific implementation steps are as follows:

[0125] Step 1: Analyze the degrees of freedom and motion state of the lower limb exoskeleton rehabilitation robot in the sagittal plane, and design the ACPG network topology to generate personalized gait trajectories with biorhythmic characteristics. Specifically, this includes:

[0126] Based on the physiological structure of the human body, and using the human body as the coordinate center, the gait motion axis can be decomposed into the sagittal, coronal, and horizontal planes. Lower limb exoskeleton robots are wearable devices used to assist patients in lower limb rehabilitation training. For safety reasons, the range of motion of the lower limbs is generally restricted to the sagittal plane during rehabilitation training, allowing the hip, knee, and ankle joints to perform flexion and extension movements within the sagittal plane. Considering the patient's limited mobility and the small range of motion of the ankle joint during walking, the ankle joint of the lower limb exoskeleton robot is usually designed as a passive joint, not participating in active movement, while the hip and knee joints act as active joints to drive the movement of the lower limbs. Therefore, the design... Figure 1 The ACPG network topology shown consists of four oscillators. Adjacent oscillators are coupled and connected by arrows. Each oscillator represents a fundamental frequency term in (1). The target trajectory of a joint consists of the superposition of the oscillator outputs within the corresponding dashed boxes, with the hip and knee joints consisting of two oscillators.

[0127] Based on Fourier series theory, any periodic function can be represented by an infinite series of sine functions. The corresponding CPG model is built based on these Fourier series terms. To reduce the complexity of the CPG model, two sine functions are used to approximate the hip and knee joint trajectory curves in gait. To achieve trajectory planning, a predetermined reference trajectory needs to be obtained first, as shown below:

[0128] (1)

[0129] Based on the given preset trajectory, the ACPG model is used to fit the above trajectory formula.

[0130] The mathematical description of the ACPG model is as follows:

[0131] (2)

[0132] in It is the first or A sine function, variable , , They represent the first The phase, amplitude, and offset of each oscillator; , , Represent the first term in equation (1) The frequency, amplitude, and offset of the state variables of a sin function; Determines the coupling strength; Determines the convergence speed; Represents scaled phase difference; discrete set Indicates the first A set of nodes from which a node receives information from its neighboring nodes; This represents the output of each node, and the trajectory of each joint can be obtained by superimposing the outputs of the nodes. The evolution of the phase over time is represented by the first equation in equation (2).

[0133] Furthermore, in this model (2), it mainly relies on and Adjusting the gait frequency and amplitude. Therefore, adjustment factors are added to the second and fourth formulas of the model, respectively. and Perform gait trajectory planning.

[0134] exist Figure 1 In the ACPG model, there are four main nodes, and the trajectory of each joint is output by combining two nodes, as shown below:

[0135] (3)

[0136] Step 2: In the ACPG model designed in Step 1, the frequency adjustment factor and amplitude adjustment factor The dynamic adjustment needs to be optimized based on real-time changes in human-machine interaction energy. Specifically, the system uses high-precision torque sensors to monitor the interaction torque at each joint during human-machine coupled motion in real time. And combined with joint angular velocity Based on the principle of function-energy conversion, the mechanical power injected into the exoskeleton system for active lower limb movement is precisely calculated. This closed-loop regulation mechanism based on biomechanical energy transfer ensures a dynamic match between the exoskeleton's assistive force and the user's movement intention.

[0137] The specific expression for calculating the energy of each joint using the interaction torque and the velocity of each joint is shown below:

[0138] (4)

[0139] By calculating the energy generated by human-computer coupling interaction, the wearer's movement intentions can be effectively captured for subsequent design of gait planning controllers.

[0140] Step 3: Construct an intermediate variable based on the weighted combination of energy generated by the work done by the interaction torques calculated in Step 2. Use this variable as input to design a gait adjustment model in conjunction with the admittance model, and output the frequency factor and amplitude factor of the adaptive compliant gait trajectory adjustment, as detailed below:

[0141] For adaptive gait adjustment strategies that emphasize human-computer interaction, the interaction effect and frequency modulation factor are adjusted. and amplitude modulation factor Based on the relationship between the mass-spring-damper type admittance model transfer function:

[0142] (5)

[0143] (6)

[0144] in, and It is the virtual inertia coefficient. and It is the virtual damping coefficient. and It is a virtual stiffness coefficient, and the admittance parameter can be adjusted according to different application scenarios and objectives. It is the energy input value of the frequency admittance model. . It is the energy input value of the amplitude admittance model. The wearer injects or dissipates human-machine interaction energy through the joints of the exoskeleton. Frequency that affects movement trajectory and amplitude coefficient Based on this characteristic, the wearer can utilize all the features of the exoskeleton. The gait trajectory of each joint is achieved by applying acceleration or deceleration torques to physically demonstrate their compliance or resistance. This effect can be used... and It is adjusted as a weighting factor for the torque of each joint during the entire gait pattern generation process.

[0145] Step 4: Based on the gait adjustment admittance interaction model designed in Step 3, and combined with CBF optimization constraint theory, the frequency and amplitude variations of the gait are constrained within a specified range, ensuring both the rhythmic gait trajectory output by the CPG network and safety. Details are as follows:

[0146] While adjusting gait frequency and amplitude based on changes in joint energy, the safety of gait adjustment also needs to be considered. In (5) and (6), and The variation is affected by the input energy and the admittance model parameters. However, the admittance model parameters are generally fixed, so and The gait primarily changes under the influence of input torque. However, the range of motion of the hip and knee joints in gait is generally limited, and the movement frequency must adapt to the wearer's own movement habits. Therefore, in gait adaptive changes, it is necessary to... and Constraints are imposed on changes in the variables. The goal of satisfying these constraints is to ensure that the variables... and We will remain within the constraints that allow it. and The constraint set is defined as follows:

[0147] (7)

[0148] To explicitly express the admittance model in the constraint set, the constraint set... Rewrite the constraints as those related to the function, with the function defined as follows:

[0149] (8)

[0150] (9)

[0151] Rewrite (7) as a constraint set constructed from functional form as follows:

[0152] (10)

[0153] To ensure forward invariance, the following condition must hold: for all and some extensions Class function , , , , After formula substitution, the following constraints can be obtained: , , , However, since the relative degrees of systems (5) and (6) are 2, there is no control input in the formula to ensure that the inequalities hold. Therefore, we define these constraints as new constraints:

[0154] (11)

[0155] (12)

[0156] (13)

[0157] To ensure the constraint set For the forward invariance of , the following conditions must be satisfied:

[0158] (14)

[0159] (15)

[0160] in To expand Class function, according to (5) and (6) we can obtain , , and The explicit form is as follows:

[0161] (16)

[0162] (17)

[0163] (18)

[0164] (19)

[0165] Based on the above analysis, the input values ​​in (5) and (6) can be solved using quadratic programming. and Make corrections to ensure the frequency factor. and amplitude factor Within safe and feasible limits, as follows:

[0166] (20)

[0167] (twenty one)

[0168] (twenty two)

[0169] in, This represents the energy value calculated based on estimated torque and joint velocity. This means that the energy value that meets the safety constraints is obtained through quadratic programming, which can guarantee the frequency factor. and amplitude factor Within safe and feasible limits.

[0170] Step 5: To verify the safety constraint optimization gait planning method for lower limb rehabilitation robots designed based on the above steps, an ACPG network model and a lower limb exoskeleton dynamics model were built in Matlab2022a / Simulink, and simulation conditions were constructed to test the proposed method. Specific settings are shown below:

[0171] In a simulation environment built using Matlab 2022a / Simulink, to simulate changes in lower limb muscle strength and observe changes in rehabilitation gait, the simulation duration was set to 10 seconds, and the patient's muscle torque was set as follows:

[0172] (twenty three)

[0173] (twenty four)

[0174] From the setting of the patient's muscle torque, such as Figure 2 As shown, it can be observed that the active torque of the patient's lower limbs increases at 2s, 4s, and 8s, and decreases at 6s. The purpose of this setting is to observe whether the proposed control algorithm will adjust the rehabilitation gait accordingly based on the increase in the patient's lower limb muscle strength.

[0175] Furthermore, the relevant parameter settings for the ACPG model are shown in Table 1 and Equation (25). The parameters for the lower limb exoskeleton dynamic model are set as follows: , , , , , , After obtaining the interaction torque, it is necessary to calculate the energy change exerted by the wearer on the joint. Then, a deformation factor is generated by constructing a virtual admittance model to achieve adaptive gait adjustment. The admittance model parameters related to the amplitude change factor are set as follows: , , The admittance model parameters related to the frequency variation factor are set as follows: , , The frequency and amplitude constraints in ACPG are respectively , , , .also, and These are the energy inputs for the two admittance models, and the weighted sums of the energies of the two joints, with the weights being: , , , Meanwhile, when using CBF to adjust the deformation factor... and When applying constraints, it is necessary to select an appropriate CBF function. In this experiment, all CBF functions selected are linear functions, i.e. , , , The PID controller serves as the underlying position controller for the lower limb exoskeleton robot. The PID parameters at the hip joint are as follows: , , The PID parameters at the knee joint are: , , .

[0176] Table 1 ACPG Model Parameters

[0177] node 1 -0.226 0.066 -0.29 2 -0.083 0.066 -0.64 3 0.413 -0.225 1.02 4 -0.240 -0.225 0.41

[0178] (25)

[0179] Simulation results are as follows Figure 3-9 As shown, the gait planning method designed in this invention can effectively generate gait joint trajectories. Figure 3 This represents the energy generated by the interactive torques acting at the hip and knee joints of the lower limbs. Figure 4 Input values ​​are used for the admittance model derived from the weighted calculation of the calculated joint energy. Observation Figure 5 and 6 It can be seen that, through the constraint of CBF, the amplitude modulation factor can be... Effective constraints are within [-0.2, 0.1] for the frequency modulation factor. The effective constraint is within [-0.1, 0.1]. Figure 7 and Figure 8It can be seen that, compared with ACPG without CBF constraint, the ACPG+CBF approach can both adjust the gait planning based on the interaction energy state and provide safety protection for gait planning. After completing the safe adaptive gait planning, gait trajectory tracking control can be achieved through PID control, such as... Figure 9 As shown.

[0180] Example 2

[0181] This embodiment relates to a gait safety planning device for a lower limb rehabilitation robot, including a memory and one or more processors. The memory stores executable code, and when the one or more processors execute the executable code, they implement a gait safety planning method for a lower limb rehabilitation robot according to the present invention.

[0182] Example 3

[0183] This embodiment relates to a computer-readable storage medium storing a program that, when executed by a processor, implements a gait safety planning method for a lower limb rehabilitation robot according to the present invention.

[0184] The above description represents only one embodiment of the present invention and does not limit the scope of its application. Those skilled in the art will recognize that the present invention can be modified and altered in many different ways. Any adjustments, substitutions, or improvements made within the core concepts and principles of the present invention should be considered part of the scope of protection of the present invention.

Claims

1. A safety constraint optimization gait planning method for lower limb rehabilitation robots, comprising the following steps: S1. Collect the gait trajectory of the hip and knee joints of one side of the human lower limb in the sagittal plane, and fit the trajectory at the hip and knee joints into mathematical expressions composed of two sine functions to obtain normal gait trajectory parameters; design an adaptive central pattern generator (ACPG) network topology structure constructed by four oscillators connected in series to adapt to the biomimetic structure of the human lower limb, and input the normal gait trajectory parameters into the ACPG model to generate the initial gait trajectory control lower limb exoskeleton rehabilitation robot; The ACPG network has four oscillators arranged in series chain form, wherein, The rhythmic signals generated by the spontaneous oscillation of the first and second oscillators represent the hip joint control trajectory, while the rhythmic signals generated by the spontaneous oscillation of the third and fourth oscillators represent the knee joint control trajectory. Four oscillators spontaneously oscillate based on the input gait trajectory parameters to generate rhythmic signals. Adjacent oscillators are coupled and coordinated to achieve smooth gait transitions. S2. Real-time acquisition of interactive torque and joint angular velocity information generated by the wearable lower limb exoskeleton rehabilitation robot, and calculation of the energy injected into the exoskeleton robot by the active movement of the human lower limb; S3. The energy generated by the interaction torque calculated in step S2 at the joint of one lower limb is weighted and combined, and the weighted energy value is used as the input value of the admittance model. The output value of the admittance model is fed back to the ACPG model as the frequency and amplitude factor of the gait trajectory, and the gait trajectory is updated online in real time. The admittance model compliantly adjusts the frequency and amplitude factors of the feedback to the ACPG model to form a personalized dynamic feedback. S4. Construct a safety constraint set for the frequency and amplitude factors of the new gait trajectory. Based on the admittance model in step S3, and combined with the safety filter characteristics of the control barrier function CBF, constrain the frequency and amplitude changes of the gait within the specified range. The safety constraint set is explicitly expressed. Construct a quadratic programming function to solve for the new frequency and amplitude factors, and feed it back into the ACPG model to generate a personalized gait trajectory with safety assurance. 2.The safety-constrained optimization gait planning method of the lower extremity rehabilitation robot according to claim 1, wherein, Step S1 specifically includes: S11 restricts the range of motion of the lower limb exoskeleton to the sagittal plane, making the hip and knee joints active joints for flexion and extension movements, and the ankle joint designed as a passive joint. It collects and fits the gait trajectory of the hip and knee joints of one side of the human lower limb in the sagittal plane. S12 constructs the ACPG network topology, which consists of four closely coupled oscillators connected in series in a chain-like manner; each oscillator can oscillate spontaneously to generate a sinusoidal signal characterizing the motion trajectory of the corresponding joint; the motion trajectories of the hip and knee joints are respectively composed of the superposition of the sinusoidal signals output by the two corresponding oscillators; S13 The first and second nodes jointly generate rhythmic signals that drive hip joint movement through internal coupling and synchronization mechanisms; the third and fourth nodes output in coordination with the same mechanism to form the knee joint movement trajectory; the four nodes cooperate with each other in the serial topology through phase coordination and amplitude modulation to finally synthesize a personalized gait trajectory that conforms to the characteristics of human movement. 3.The safety-constrained optimization gait planning method of the lower extremity rehabilitation robot according to claim 2, wherein, Step S12 specifically includes: S121. Design a network topology consisting of 4 oscillators, with adjacent oscillators connected by arrows to form a coupling relationship; S122. Each oscillator corresponds to a fundamental frequency term in a Fourier series; S123. The target trajectory of the hip joint is formed by the superposition of the outputs of the first oscillator and the second oscillator, and the target trajectory of the knee joint is formed by the superposition of the outputs of the third oscillator and the fourth oscillator; Step S13 specifically includes: S131. Based on Fourier series theory, the hip / knee joint motion curve is approximated using the preset reference trajectory defined by equation (1): （1） S132. Establish the ACPG dynamic model as shown in equation (2): （2） in It is the first or A sine function, variable , , They represent the first The phase, amplitude, and offset of each oscillator; , , Represent the first term in equation (1) respectively The frequency, amplitude, and offset of the state variables of a sin function; Determines the coupling strength; Determines the convergence speed; Represents scaled phase difference; discrete set Indicates the first A set of nodes from which a node receives information from its neighboring nodes; The output of each node is represented by the superposition of the node outputs; the evolution of the phase over time is represented by the first equation in equation (2); S133. Joint trajectory output satisfies equation (3): （3） S134. By adjusting factors and Adjusting the frequency and amplitude of the gait; therefore, adjustment factors are added to the second and fourth formulas of equation (2) respectively. and Perform gait trajectory planning.

4. The safety constraint optimization gait planning method for lower limb rehabilitation robots as described in claim 1, characterized in that, Step S2 specifically includes: S21. Real-time monitoring of joint interaction torques in human-machine coupled motion. With joint angular velocity ; S22. Calculate the mechanical power of the human lower limb active injection exoskeleton system based on the function-energy conversion principle, and calculate the energy transfer value of a single joint according to formula (4): （4） S23. By combining multi-joint energy weights, the wearer's movement intention is captured based on the calculated energy value generated by human-computer coupling interaction.

5. The safety constraint optimization gait planning method for lower limb rehabilitation robots as described in claim 1, characterized in that, Step S3 specifically includes: S31 Constructs intermediate variables based on human-computer interaction energy; S32 Establish the mass-spring-damper type admittance transfer function; S33 outputs the frequency factor and amplitude factor for adaptive gait adjustment.

6. The safety constraint optimization gait planning method for a lower limb rehabilitation robot as described in claim 5, characterized in that, Step S31 specifically includes: S311. Use energy-weighted combination and Construct intermediate variables, and It is adjusted as a weighting factor for the torque of each joint during the entire gait pattern generation process; S312. Define frequency-dependent energy input , It is the energy input value of the frequency admittance model; S313. Define amplitude-dependent energy input , It is the energy input value of the amplitude admittance model. Step S32 specifically includes: S321. Establish the frequency admittance model: （5） S322. Establish the magnitude admittance model: （6） in, and It is the virtual inertia coefficient. and It is the virtual damping coefficient. and It is a virtual stiffness coefficient, and the admittance parameter can be adjusted according to different application scenarios and objectives; The implementation of step S33 includes: through human-computer interaction energy Frequency factors affecting motion trajectory and amplitude factor Based on this characteristic, the wearer utilizes all the features of the exoskeleton. The gait trajectory of each joint is achieved by applying acceleration or deceleration torque to physically demonstrate their compliance or resistance.

7. The safety constraint optimization gait planning method for lower limb rehabilitation robots as described in claim 1, characterized in that, Step S4 specifically includes: S41. Define the gait frequency modulation factor and amplitude modulation factor The set of safety constraints; S42. Construct forward invariant constraints based on the control barrier function (CBF); S43. Solve for the safe energy input value using quadratic programming.

8. The safety constraint optimization gait planning method for lower limb rehabilitation robots as described in claim 7, characterized in that, Step S41 specifically includes: S411. Establish the constraint set Limited frequency factor and amplitude factor Feasible range: （7） S412 Construct constraint functions (8) and (9): （8） （9） S413. Transform the constraint set into function form: （10） Step S42 specifically includes: S421. Derive new constraints through relative degree analysis: （11） （12） （13） S422. Design forward invariance conditions (14)-(15): （14） （15） S423. Calculate the relevant derivative terms (16)-(19): （16） （17） （18） （19） Step S43 specifically includes: Constructing quadratic programming problems (20)-(22): （20） （21） （20） in, This represents the energy value calculated based on estimated torque and joint velocity. This means that the energy value that meets the safety constraints is obtained through quadratic programming, which can guarantee the frequency factor. and amplitude factor Within safe and feasible limits.

9. The safety constraint optimization gait planning method for a lower limb rehabilitation robot as described in claim 1, characterized in that, The method also includes step S5: building an ACPG network model and a lower limb exoskeleton dynamics model in Matlab2022a / Simulink, and constructing simulation conditions to verify and test the proposed method.

10. A safety constraint optimization gait planning device for a lower limb rehabilitation robot, characterized in that, The device includes a memory and one or more processors, wherein the memory stores executable code, and the one or more processors execute the executable code to implement the safety constraint optimization gait planning method for the lower limb rehabilitation robot according to any one of claims 1-9.

Images