Joint angle vector generation method and gait control method

Abstract

The application relates to the technical field of rehabilitation medical robots, and particularly provides a joint angle vector generation method and a gait control method. The generation method comprises the following steps: acquiring a first gait phase and multi-sensor data at a current moment; the first gait phase is a gait phase of a lower-limb exoskeleton robot at a previous moment, and the multi-sensor data comprises multi-modal motion data and visual data; a second gait phase is predicted according to the first gait phase and the multi-modal motion data; the second gait phase is a gait phase of the lower-limb exoskeleton robot at the current moment; a reference joint angle vector and a time scaling factor of a pre-optimized adaptive joint angle vector generator are adjusted according to the visual data, and then an expected joint angle vector is generated according to the second gait phase by using the adjusted adaptive joint angle vector generator; the method can generate a joint angle vector which is suitable for a current environment and has high gait bionics.

Description

Technical Field

[0001] This application relates to the field of rehabilitation medical robot technology, and more specifically, to a method for generating joint angle vectors and a gait control method. Background Technology

[0002] With the accelerating aging of the global population and the continued rise in the incidence of neurological diseases such as stroke and spinal cord injury, the number of patients with lower limb motor dysfunction due to neurological damage has increased significantly. For these patients, regaining independent walking ability is a core requirement for achieving self-care and reintegration into society. Traditional rehabilitation training heavily relies on manual assistance from therapists, which presents inherent bottlenecks such as difficulty in standardizing treatment intensity, high labor costs, limited training efficiency, and difficulty in quantifying and evaluating rehabilitation effects.

[0003] Lower limb exoskeleton robots, as a product of the deep integration of modern rehabilitation medicine and robotics, provide support and assistance to patients through mechanical structures, helping them complete standardized walking training and offering an effective technical solution to the aforementioned problems. However, current technologies still face many challenges in practical applications: First, existing systems rely on single sensor data for simple threshold judgments, failing to achieve accurate phase estimation; second, existing systems use fixed-parameter control models, lacking the ability to deeply learn and adaptively adjust to the gait characteristics of healthy individuals; and finally, existing systems cannot dynamically adjust joint angle vectors according to the surrounding environment.

[0004] If the above problems are not solved, lower limb exoskeleton robots will be limited to operating in structured indoor environments and will not be able to meet the rehabilitation needs of patients in real-life scenarios. The lack of environmental perception makes the device slow to respond to changes in terrain, increasing the risk of gait abnormalities. Insufficient gait biomimicry may induce non-physiological movement patterns, which is not conducive to the re-establishment of damaged neural pathways, ultimately resulting in limited rehabilitation training effects and a narrow range of applications.

[0005] There is currently no effective technical solution to the above problems. Summary of the Invention

[0006] The purpose of this application is to provide a joint angle vector generation method and a gait control method that can generate joint angle vectors that are adapted to the current environment and have high gait biomimicry.

[0007] In a first aspect, this application provides a method for generating joint angle vectors, applied in a lower limb exoskeleton robot, which includes the following steps:

[0008] A1. Acquire the first gait phase and multi-sensor data at the current moment; the first gait phase is the gait phase of the lower limb exoskeleton robot at the previous moment, and the multi-sensor data includes multimodal motion data and visual data;

[0009] A2. Predict the second gait phase based on the first gait phase and multimodal motion data; the second gait phase is the gait phase of the lower limb exoskeleton robot at the current moment;

[0010] A3. Adjust the reference joint angle vector and time scaling factor of the pre-optimized adaptive joint angle vector generator based on visual data, and then use the adjusted adaptive joint angle vector generator to generate the desired joint angle vector based on the second gait phase.

[0011] The adaptive joint angle vector generator adopts a dynamic motion primitive architecture. The pre-optimization weight process of the adaptive joint angle vector generator includes:

[0012] S1. Collect teaching gait data of healthy individuals in different environments;

[0013] S2. Generate the optimal weight reassembly of the adaptive joint angle vector generator based on all taught gait data. The optimal weight reassembly includes the optimal weights corresponding to different lower limb joints.

[0014] This application provides a joint angle vector generation method. By pre-generating an adaptive joint angle vector generator using the optimal weight reorganization of gait data from healthy individuals in different environments, the adaptive joint angle vector generator possesses deep learning and adaptive adjustment capabilities based on the gait characteristics of healthy individuals. Phase prediction based on multi-sensor data is achieved by predicting the second gait phase using first gait data and multimodal motion data, effectively improving the accuracy and reliability of phase estimation. Furthermore, by first adjusting the reference joint velocity vector and time scaling factor of the adaptive joint angle vector generator based on visual data, the parameters of the adaptive joint angle vector generator are adjusted according to the surrounding environment. Then, the adjusted adaptive joint angle vector generator generates the desired joint angle based on the second gait phase, achieving dynamic adjustment of the desired joint angle vector according to the surrounding environment. Therefore, this application provides a method that can generate joint angle vectors that are adaptable to the current environment and have high gait biomimicry, effectively meeting the rehabilitation needs of patients in real-life scenarios and avoiding the problems of increased gait abnormalities due to a lack of environmental awareness, which can induce non-physiological movement patterns and hinder the re-establishment of damaged neural pathways. This effectively improves the rehabilitation training effect and expands the application scope of rehabilitation training.

[0015] Optionally, step A2 includes:

[0016] A21. The first initial gait phase is generated based on the first step gait phase using the state transition equation of the extended Kalman filter; the first initial gait phase is the predicted value of the gait phase at the current moment generated based on the first step gait phase.

[0017] A22. The observation equation of the extended Kalman filter is used to generate a second initial gait phase based on the multimodal motion data; the second initial gait phase is the observed value of the gait phase at the current moment generated based on the phase of the multimodal motion data.

[0018] A23. The first initial gait phase is corrected by using the extended Kalman filter based on the second initial gait phase to obtain the second gait phase.

[0019] Optionally, the multimodal motion data includes multiple motion parameters, and step A22 includes:

[0020] A221. For each motion parameter, the observation equation of the extended Kalman filter is used to generate the third initial gait phase corresponding to that motion parameter.

[0021] A222. Obtain the second initial gait phase based on all third initial gait phases.

[0022] Optionally, the motion parameters are heel pressure or knee angle, and the formula for calculating the third initial gait phase corresponding to heel pressure is as follows:

[0023] ;

[0024] in, Indicates heel pressure. This indicates the preset coefficient corresponding to heel pressure. This indicates the third initial gait phase corresponding to heel pressure;

[0025] The formula for calculating the third initial gait phase corresponding to the knee joint angle is as follows:

[0026] ;

[0027] in, Indicates the knee joint angle. This represents the preset coefficient corresponding to the knee joint angle. This represents the third initial gait phase corresponding to the knee joint angle. This indicates the preset phase offset corresponding to the knee joint angle.

[0028] Alternatively, the formula for calculating the desired joint angle vector is as follows:

[0029] ;

[0030] Where t represents time. Let z(t) represent the time scaling factor, and z(t) represent the auxiliary vector. The derivative of the auxiliary vector, and Both represent gain constants. This represents the desired joint angle vector. Let denote the derivative of the desired joint angle vector, and g denote the reference joint angle vector. Indicates the second gait phase. This represents the nonlinear forcing term corresponding to the second step phase;

[0031] The formula for calculating the nonlinear forcing term is shown below:

[0032] ;

[0033] in, Let N represent the Gaussian radial basis functions, and let N be the number of Gaussian radial basis functions. This represents the width of the i-th Gaussian radial basis function. This represents the center of the i-th Gaussian radial basis function. This represents the weight corresponding to the i-th Gaussian radial basis function.

[0034] Optionally, step S2 includes:

[0035] S21. Optimal weight reorganization of the adaptive joint angle vector generator based on all taught gait data using a local weighted regression algorithm.

[0036] Secondly, this application also provides a gait control method for use in a lower limb exoskeleton robot, which includes the following steps:

[0037] B1. Acquire the first gait phase and multi-sensor data at the current moment; the first gait phase is the gait phase of the lower limb exoskeleton robot at the previous moment, and the multi-sensor data includes multimodal motion data and visual data;

[0038] B2. Predict the second gait phase based on the first gait phase and multimodal motion data; the second gait phase is the gait phase of the lower limb exoskeleton robot at the current moment;

[0039] B3. Adjust the reference joint angle vector and time scaling factor of the pre-optimized adaptive joint angle vector generator based on visual data, and then use the adjusted adaptive joint angle vector generator to generate the desired joint angle vector based on the second gait phase.

[0040] B4. Determine the current gait phase based on the second gait phase and multimodal motion data;

[0041] B5. Obtain the desired joint torque based on the desired joint angle vector;

[0042] B6. While keeping the desired joint torque constant, adjust the stiffness and damping of the lower limb exoskeleton robot according to the current gait stage;

[0043] The adaptive joint angle vector generator adopts a dynamic motion primitive architecture. The pre-optimization weight process of the adaptive joint angle vector generator includes:

[0044] S1. Collect teaching gait data of healthy individuals in different environments;

[0045] S2. Generate the optimal weight reassembly of the adaptive joint angle vector generator based on all taught gait data. The optimal weight reassembly includes the optimal weights corresponding to different lower limb joints.

[0046] This application provides a gait control method that utilizes optimal weight reorganization of gait data from healthy individuals in different environments to pre-generate an adaptive joint angle vector generator. This enables the generator to possess deep learning and adaptive adjustment capabilities for the gait characteristics of healthy individuals. Phase prediction based on multi-sensor data is achieved by predicting the second gait phase using first gait data and multimodal motion data, effectively improving the accuracy and reliability of phase estimation. Furthermore, the method first adjusts the reference joint velocity vector and time scaling factor of the adaptive joint angle vector generator based on visual data, allowing for parameter adjustment according to the surrounding environment. Then, the adjusted adaptive joint angle vector generator generates the desired joint angle based on the second gait phase, dynamically adjusting the desired joint angle vector according to the surrounding environment. Therefore, this application provides a method that can generate joint angle vectors that adapt to the current environment and exhibit high gait biomimicry, effectively meeting the rehabilitation needs of patients in real-life scenarios and avoiding problems such as increased gait abnormalities due to a lack of environmental awareness, which can induce non-physiological movement patterns and hinder the re-establishment of damaged neural pathways. This effectively improves rehabilitation training outcomes and expands the application scope of rehabilitation training.

[0047] Optionally, step B6 includes:

[0048] B61. Obtain the torque for human-computer interaction;

[0049] B62. While keeping the desired joint torque constant, adjust the stiffness and damping of the lower limb exoskeleton robot according to the current gait stage and human-machine interaction torque.

[0050] Alternatively, the formula for calculating the desired joint torque is as follows:

[0051] ;

[0052] in, Let J represent the desired joint torque, and J represent the Jacobian matrix of the lower limb skeletal robot. This represents the transpose of the Jacobian matrix of a lower limb skeletal robot. This represents the preset virtual stiffness. This represents the preset virtual damping. Let represent the desired joint angle vector, and q(t) represent the actual joint angle vector. This represents the derivative of the desired joint angle vector. The derivative of the actual joint angle vector. This indicates the preset gravity compensation item.

[0053] Optionally, step B4 includes:

[0054] B41. The gait phase corresponding to the preset phase mapping interval to which the second gait phase belongs and the preset conditions satisfied by the multimodal motion data is taken as the current gait phase.

[0055] As can be seen from the above, the joint angle vector generation method and gait control method provided in this application, by pre-generating the optimal weight reorganization of the adaptive joint angle vector generator using the taught gait data of healthy individuals in different environments, enables the adaptive joint angle vector generator to possess the ability to deeply learn and adaptively adjust to the gait characteristics of healthy individuals. Phase prediction based on multi-sensor data is achieved by predicting the second gait phase according to the first gait data and multimodal motion data, thereby effectively improving the accuracy and reliability of phase estimation. Furthermore, by first adjusting the reference joint velocity vector and time scaling factor of the adaptive joint angle vector generator based on visual data, the root... The parameters of the adaptive joint angle generator are adjusted according to the surrounding environment. Then, the adjusted adaptive joint angle vector generator generates the desired joint angle based on the second gait phase. This achieves dynamic adjustment of the desired joint angle vector according to the surrounding environment. Therefore, this application provides a method to generate joint angle vectors that are adapted to the current environment and have high gait biomimicry. This effectively meets the rehabilitation needs of patients in real-life scenarios and avoids the problem of increased gait abnormalities due to lack of environmental awareness, which can induce non-physiological movement patterns and affect the re-establishment of damaged neural pathways. This effectively improves the rehabilitation training effect and expands the application scope of rehabilitation training. Attached Figure Description

[0056] Figure 1 A flowchart illustrating a joint angle vector generation method provided in this application embodiment.

[0057] Figure 2 This is a flowchart of a gait control method provided in an embodiment of this application. Detailed Implementation

[0058] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0059] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0060] In traditional lower limb rehabilitation exoskeleton robot systems, the lack of environmental perception capabilities prevents the devices from adapting to complex environments and terrains. Specifically, this manifests as a lack of a multi-sensor fusion perception framework, relying solely on joint encoders for joint angle vector tracking, which fails to accurately estimate gait phase and identify environmental features in real time. Furthermore, insufficient gait biomimicry is reflected in the control strategy being based on predefined fixed joint angle curves, lacking dynamic adaptive modeling. This results in stiff gait joint angle vectors that do not conform to biomechanical principles, affecting the continuity and naturalness of the training process. Specifically, the lack of environmental perception makes it difficult for the device to distinguish between flat surfaces and obstacles, while insufficient gait biomimicry leads to joint movements lacking the flexibility of a healthy person's gait, resulting in a disconnect between the training scenario and real-life scenarios, thus limiting the effectiveness of rehabilitation training.

[0061] For example, in home rehabilitation scenarios, when patients use lower limb exoskeleton robots for gait training and encounter carpet edges or slight slopes, the device, operating solely on joint encoder data, cannot recognize terrain changes through plantar pressure sensors or visual data, leading to gait phase prediction errors. Furthermore, the predefined joint angle vectors are not dynamically adjusted based on environmental information, causing the exoskeleton to continue executing a fixed movement pattern on flat surfaces. This results in gait interruptions or instability when the patient crosses obstacles, forcing the training process to stop. Therefore, the device cannot maintain continuous gait training in unstructured environments, preventing patients from obtaining a rehabilitation experience that matches their daily activities, directly impacting the achievement of training goals.

[0062] If the above problems are not solved, lower limb exoskeleton robots will be limited to operating in structured indoor environments and will not be able to meet the rehabilitation needs of patients in real-life scenarios. The lack of environmental perception makes the device slow to respond to changes in terrain, increasing the risk of gait abnormalities. Insufficient gait biomimicry may induce non-physiological movement patterns, which is not conducive to the re-establishment of damaged neural pathways, ultimately resulting in limited rehabilitation training effects and a narrow range of applications.

[0063] In this regard, firstly, this application provides a method for generating joint angle vectors, applied in a lower limb exoskeleton robot, which includes the following steps:

[0064] A1. Acquire the first gait phase and multi-sensor data at the current moment; the first gait phase is the gait phase of the lower limb exoskeleton robot at the previous moment, and the multi-sensor data includes multimodal motion data and visual data;

[0065] A2. Predict the second gait phase based on the first gait phase and multimodal motion data; the second gait phase is the gait phase of the lower limb exoskeleton robot at the current moment;

[0066] A3. Adjust the reference joint angle vector and time scaling factor of the pre-optimized adaptive joint angle vector generator based on visual data, and then use the adjusted adaptive joint angle vector generator to generate the desired joint angle vector based on the second gait phase.

[0067] The adaptive joint angle vector generator adopts a dynamic motion primitive architecture. The pre-optimization weight process of the adaptive joint angle vector generator includes:

[0068] S1. Collect teaching gait data of healthy individuals in different environments;

[0069] S2. Generate the optimal weight reassembly of the adaptive joint angle vector generator based on all taught gait data. The optimal weight reassembly includes the optimal weights corresponding to different lower limb joints.

[0070] To facilitate understanding, some key terms in this application are explained below. The first step gait phase in this embodiment refers to the position of the lower limb exoskeleton robot in the previous gait cycle. This first step gait phase can be represented as a percentage from 0% to 100%, where 0% represents the beginning of a gait cycle (e.g., heel strike) and 100% represents the end of a gait cycle. This embodiment essentially uses the first step gait phase to provide a historical reference for predicting the current gait. The multi-sensor data in this embodiment refers to the comprehensive information acquired from multiple sensors at the current moment. This multi-sensor data includes multimodal motion data and visual data. The multimodal motion data typically comes from an inertial measurement unit (IMU) or plantar pressure sensors to capture the motion state and force conditions of the lower limbs. The visual data comes from visual sensors such as cameras to perceive environmental features and terrain information. This embodiment provides a more comprehensive and accurate environmental perception capability by acquiring multi-sensor data at the current moment. The second gait phase in this embodiment refers to the position of the lower limb exoskeleton robot in the gait cycle at the current moment. This second gait phase is predicted based on the gait phase at the previous moment and the current multimodal motion data. This second gait phase is a key input for generating the desired joint angle vector. The adaptive joint angle vector generator in this embodiment is a controller based on a Dynamic Motion Primitive (DMP) architecture. Its core function is to generate a desired joint angle vector that conforms to biomechanical characteristics based on the input gait phase. This generator has adaptive capabilities and can adjust its output according to environmental and task requirements. The reference joint angle vector in this embodiment refers to the baseline joint angle vector used by the adaptive joint angle vector generator when generating the desired joint angle vector. This vector typically represents an ideal, healthy gait pattern and can be adjusted based on visual data to adapt to the current environment of the lower limb exoskeleton robot. The time scaling factor in this embodiment refers to a parameter used to adjust the duration of the gait cycle. This embodiment can change the speed of the joint angle vector by adjusting the time scaling factor to adapt to different walking speeds or environmental requirements. The optimal weight reassembly in this embodiment refers to the set of parameters used to adjust the behavior of dynamic motion primitives in the adaptive joint angle vector generator. These weights are obtained by optimizing and learning from the gait data of healthy people in different environments to ensure that the generated joint angle vector has good biomimicry and adaptability. The optimal weight reassembly includes the optimal weights corresponding to different lower limb joints (such as hip joints and knee joints) so that the movement of each joint can be finely controlled.

[0071] This application provides a joint angle vector generation method for use in lower limb exoskeleton robots, aiming to solve the problems of lack of environmental perception and insufficient gait biomimicry in traditional exoskeleton robots when generating gait in complex environments.

[0072] First, various methods can be used to acquire the first gait phase and the multi-sensor data at the current moment. For example, the acquisition method of the first gait phase is the same as that of the second gait phase. That is, in this embodiment, the second gait phase is used as the first gait phase at the next moment. The multi-sensor data can include information collected by plantar pressure sensors, inertial measurement units (IMUs), and vision sensors (such as depth cameras or RGB cameras). Specifically, plantar pressure sensors can provide pressure distribution data of the foot in contact with the ground, IMUs can provide posture, angular velocity, and acceleration data of the exoskeleton joints, and vision sensors can capture images or depth information of the environment. These data can be acquired independently or collected by a data bus or wireless communication to a central processing unit for unified processing.

[0073] Secondly, different prediction models can be used to predict the second gait phase based on the first gait phase and multimodal motion data. For example, a simple linear regression model can be used to predict the gait phase at the next moment based on the changing trends of historical gait phases and current multimodal motion data. Another approach is to use a rule-based finite state machine to switch gait phases based on preset gait events (such as heel strike and toe lift) and multimodal motion data, thereby inferring the second gait phase. Machine learning-based methods can also be used to train a model to learn the complex nonlinear relationship between gait phase and multimodal motion data, thereby enabling prediction.

[0074] Subsequently, the reference joint angle vector and time scaling factor of the pre-optimized adaptive joint angle vector generator are adjusted based on visual data. Then, the adjusted adaptive joint angle vector generator is used to generate the desired joint angle vector based on the second gait phase. There are several ways to implement this. For example, visual data can be used to identify the type of the current environment, such as flat ground, uphill, downhill, or stairs. Based on the identified environment type, a suitable reference joint angle vector can be selected from a pre-set reference joint angle vector library. Simultaneously, the time scaling factor can also be adjusted according to the environment type; for example, gait speed may need to be slowed down when going uphill, thus reducing the time scaling factor. The adaptive joint angle vector generator in this embodiment adopts a Dynamic Motion Primitive (DMP) architecture. Its pre-optimization weight process includes: first, collecting teaching gait data of healthy individuals in different environments. This teaching data can include walking data of healthy individuals in various terrains such as flat ground, uphill, downhill, and stairs; then, generating an optimal weight reassembly for the adaptive joint angle vector generator based on all teaching gait data. This optimal weight reassembly includes the optimal weights corresponding to different lower limb joints; for example, the weights for the hip and knee joints can be optimized separately. When generating the desired joint angle vector, the adaptive joint angle vector generator combines the adjusted reference joint angle vector, the time scaling factor, and the currently predicted second gait phase to output a smooth and environmentally adapted desired joint angle vector through a dynamic motion primitive architecture.

[0075] The following example provides a more detailed explanation of the above technical solution: Suppose user A is a patient with lower limb motor dysfunction and is using a lower limb exoskeleton robot for rehabilitation training. This exoskeleton robot needs to assist user A in transitioning from flat ground to a slightly uphill section in an indoor environment. First, while user A is walking on flat ground, the lower limb exoskeleton robot continuously acquires the first step phase and multi-sensor data at the current moment. For example, at a certain moment, the exoskeleton robot records the first step phase from the previous moment as 50% (i.e., the midpoint of the gait cycle). Simultaneously, it acquires multimodal motion data and visual data at the current moment through plantar pressure sensors, an inertial measurement unit (IMU), and a vision sensor. The multimodal motion data may include heel strike pressure, knee joint angle, hip joint angular velocity, etc., while the visual data may capture image information of the gradually rising ground ahead. Next, based on the acquired first gait phase and multimodal motion data, the lower limb exoskeleton robot predicts the second gait phase. For example, using a pre-trained gait phase prediction model, combined with 50% of the gait phase from the previous moment and multimodal motion data such as plantar pressure and knee angle at the current moment, the robot predicts the second gait phase at the current moment to be 52%. This prediction process considers the continuity of gait and the dynamics of motion. Subsequently, the exoskeleton robot adjusts the parameters of the adaptive joint angle vector generator using visual data. When the visual sensor detects an uphill section ahead, the system, based on this visual information, reduces the reference joint angle vector of the adaptive joint angle vector generator to 0.9 times the original reference joint angle vector and increases the time scaling factor of the adaptive joint angle vector generator to 1.1 times the original time scaling factor. This adaptive joint angle vector generator, during pre-optimization of weights, has already collected gait data from healthy individuals in different environments such as flat ground and uphill slopes, and generated an optimal weighted reassembly including different lower limb joints such as the hip and knee joints based on this data. Using these adjusted reference joint angle vectors and time scaling factors, combined with the currently predicted second gait phase (52%), the adaptive joint angle vector generator generates a series of desired joint angle vectors. These vectors guide the exoskeleton robot's hip and knee joints to move smoothly, naturally, and adaptably to uphill terrain. Thus, the exoskeleton robot can drive its joints to perform corresponding movements based on the generated desired joint angle vectors, thereby assisting user A in smoothly and naturally transitioning from flat ground to an uphill section. The entire process achieves dynamic adaptation from environmental perception to joint angle vector generation, solving the problem of stiff and unnatural gait in traditional lower limb exoskeleton robots in complex environments.

[0076] This application significantly improves the environmental perception and gait biomimicry of lower limb exoskeleton robots in complex environments by fusing multi-sensor data and a dynamic motion primitive architecture. Traditional exoskeleton robots typically rely on pre-programmed fixed gait patterns, which cannot adjust their gait in real time when facing complex terrain such as uphill and downhill slopes, resulting in stiff, unnatural gait and potentially causing falls. For example, in the scenario described above where user A transitions from flat ground to an uphill slope, existing fixed gait patterns may not be able to adapt to the joint angle and speed changes required for uphill walking, making it difficult for user A to walk. This application, by acquiring multi-sensor data, particularly incorporating visual data, enables the exoskeleton robot to perceive environmental changes in real time, such as identifying uphill sections ahead. This contrasts sharply with existing technologies that lack an effective multi-sensor fusion framework and rely solely on joint encoders for simple angle tracking. By combining the first gait phase and multimodal motion data to predict the second gait phase, this application can more accurately grasp the position in the current gait cycle, providing precise input for subsequent gait generation. Furthermore, this application utilizes visual data to dynamically adjust the reference joint angle vector and time scaling factor of the adaptive joint angle vector generator. This dynamic adjustment mechanism enables the exoskeleton robot to generate adaptive gait based on the actual environment (such as uphill), rather than a rigid preset pattern. The adaptive joint angle vector generator adopts a dynamic motion primitive (DMP) architecture and optimizes the weights by collecting taught gait data from healthy individuals in different environments, ensuring that the generated desired joint angle vector has a high degree of biomimicry and naturalness. Compared with existing methods that simply average or fit healthy individuals' gait data, lacking dynamic adaptive gait modeling, this approach better activates damaged neural pathways in patients and avoids the generation of abnormal gait patterns. Therefore, the technical solution of this application can provide patients with a safer, more efficient, and personalized rehabilitation training experience, significantly improving the practicality and rehabilitation effect of lower limb exoskeleton robots.

[0077] Therefore, the joint angle vector method provided in this application enables the adaptive joint angle vector generator to possess deep learning and adaptive adjustment capabilities for the gait characteristics of healthy individuals by pre-generating an adaptive joint angle vector generator using the optimal weight reorganization of gait data from healthy individuals in different environments. It achieves phase prediction based on multi-sensor data by predicting the second gait phase based on the first gait data and multimodal motion data, effectively improving the accuracy and reliability of phase estimation. Furthermore, it adjusts the reference joint velocity vector and time scaling factor of the adaptive joint angle vector generator based on visual data to adjust the parameters of the adaptive joint angle vector generator according to the surrounding environment. Then, it uses the adjusted adaptive joint angle vector generator to generate the desired joint angle based on the second gait phase, achieving dynamic adjustment of the desired joint angle vector according to the surrounding environment. Therefore, this application provides a method to generate joint angle vectors that are adaptable to the current environment and have high gait biomimicry, effectively meeting the rehabilitation needs of patients in real-life scenarios and avoiding the problems of increased gait abnormalities due to a lack of environmental awareness, which can induce non-physiological movement patterns and hinder the re-establishment of damaged neural pathways. This effectively improves the rehabilitation training effect and expands the application scope of rehabilitation training.

[0078] In some preferred embodiments, step A2 includes:

[0079] A21. The first initial gait phase is generated based on the first step gait phase using the state transition equation of the extended Kalman filter; the first initial gait phase is the predicted value of the gait phase at the current moment generated based on the first step gait phase.

[0080] A22. The observation equation of the extended Kalman filter is used to generate a second initial gait phase based on the multimodal motion data; the second initial gait phase is the observed value of the gait phase at the current moment generated based on the phase of the multimodal motion data.

[0081] A23. The first initial gait phase is corrected by using the extended Kalman filter based on the second initial gait phase to obtain the second gait phase.

[0082] The Extended Kalman Filter (EKF) in this embodiment is a state estimation algorithm applicable to nonlinear systems. It estimates the system state through an iterative prediction and update process. In step A21, the EKF state transition equation is first used to predict the current gait state based on the first step phase of the lower limb exoskeleton robot at the previous moment, combined with the system's dynamic model, thereby generating a first initial gait phase. This first initial gait phase represents a priori estimation based on the system's own motion laws. Subsequently, in step A22, the EKF observation equation is used to observe the current gait state based on multimodal motion data from the multi-sensor data acquired at the current moment, thereby generating a second initial gait phase. This second initial gait phase reflects the observation results based on the actual measurement data. Finally, in step A23, the EKF fuses and corrects the first initial gait phase (predicted value) and the second initial gait phase (observed value). Specifically, step A23 calculates the Kalman gain and performs a weighted average of the predicted and observed values ​​to obtain an optimal gait phase estimate (i.e., the second gait phase) that has been freed from noise. It should be understood that the extended Kalman filter used in this application is prior art, and its working principle will not be discussed in detail here.

[0083] This embodiment effectively addresses the noise interference and model uncertainty issues inherent in traditional gait phase prediction by introducing an extended Kalman filter. Specifically, the extended Kalman filter first uses the state transition equation to predict the gait phase at the current moment based on the first gait phase at the previous moment, generating a first initial gait phase, which provides a prior estimate based on the system's dynamic model. Subsequently, through the observation equation, combined with the multimodal motion data at the current moment, the gait phase is observed to generate a second initial gait phase. Finally, the extended Kalman filter weighted and fused the predicted value (first initial gait phase) and the observed value (second initial gait phase) to correct the prediction result. This iterative prediction-correction process allows the system to adjust the gait phase estimate in real time based on the latest sensor data, thus achieving high-precision gait phase prediction results even in the presence of noise and uncertainty. Therefore, accurate gait phase identification and tracking can be ensured even in complex motion environments or when sensor data quality is poor. Through the above technical solution, this application utilizes extended Kalman filtering for gait phase prediction and correction, significantly improving the accuracy and robustness of gait phase estimation. Compared to simple model-based or purely sensor-data-based prediction methods, Extended Kalman Filtering (EKF) effectively suppresses the influence of sensor noise and provides optimal estimation of the system state, ensuring that the lower limb exoskeleton robot can more accurately sense and track the user's gait cycle. This not only improves the real-time performance and accuracy of gait assistance in exoskeleton robots but also enhances the system's stability and safety, providing users with a more natural and comfortable wearing experience. It effectively avoids problems such as lag or over-assistance caused by inaccurate gait phase prediction.

[0084] In some preferred embodiments, the multimodal motion data includes multiple motion parameters, and step A22 includes:

[0085] A221. For each motion parameter, the observation equation of the extended Kalman filter is used to generate the third initial gait phase corresponding to that motion parameter.

[0086] A222. Obtain the second initial gait phase based on all third initial gait phases.

[0087] The multimodal motion data in this embodiment refers to the set of motion parameters collected by different types of sensors on the lower limb exoskeleton robot. These motion parameters may include, but are not limited to, heel and toe pressure collected by plantar pressure sensors, angular velocity and acceleration provided by inertial measurement units (IMUs), knee and hip joint angles measured by joint encoders, and muscle activity signals measured by electromyography (EMG) sensors. First, for each motion parameter, the system uses the observation equation of Kalman filtering to generate a third initial gait phase corresponding to that motion parameter. This step aims to map and convert each independent motion parameter observation value into a preliminary gait phase estimate associated with that parameter through the observation model of Kalman filtering, allowing the system to independently evaluate the contribution of each sensor data and prepare for subsequent fusion. For example, for heel pressure data, an observation equation can be established to correlate heel pressure values ​​with gait phase (e.g., using a nonlinear function to describe the periodic variation of heel pressure with gait phase). Similarly, for knee angle data, an observation equation can be constructed to correlate knee angle values ​​with gait phase (e.g., using a nonlinear function to describe the periodic variation of knee angle values ​​with gait phase). Subsequently, the system obtains a second initial gait phase based on all third initial gait phases to effectively fuse the preliminary gait phase estimates obtained from each independent motion parameter, thereby obtaining a more comprehensive and robust second initial gait phase. Specifically, the system can use the average of all third initial gait phases as the second initial gait phase, or it can use a weighted average method, assigning different weights based on the reliability or importance of each motion parameter, and then performing a weighted average on all third initial gait phases to obtain the final second initial gait phase.

[0088] This embodiment effectively solves the problem of inaccurate multi-parameter data fusion by decomposing multimodal motion data into independent motion parameters and then fusing them. Specifically, when predicting the second gait phase, the system no longer directly inputs the multimodal motion data as a whole into the Kalman filter observation equation. Instead, for each independent motion parameter, such as heel pressure or knee angle, a third initial gait phase corresponding to that motion parameter is generated using the Kalman filter observation equation. This divide-and-conquer approach ensures that the characteristics of each motion parameter can be independently and accurately modeled and utilized, avoiding interference from dimensional differences, noise characteristic differences, or nonlinear relationships that may exist between different types of sensor data on the overall observation model, thereby enhancing the relevance and reliability of data processing. Subsequently, the system integrates all these independently generated third initial gait phases to obtain the final second initial gait phase. This integration process can effectively fuse information from different sensors, compensating for the limitations or errors that may exist in a single sensor. For example, when the data from one sensor is temporarily unreliable, the information from other sensors can still provide effective phase estimation. In this way, the proposed solution achieves effective fusion of multi-source data, resulting in more robust and accurate gait phase prediction. This embodiment avoids direct conflicts and mutual interference between different sensor data by independently generating a third initial gait phase for each motion parameter, allowing for more precise utilization of the observation information for each parameter. Furthermore, this embodiment achieves effective fusion of multimodal motion data by synthesizing a second initial gait phase based on all third initial gait phases, resulting in a more robust and accurate second initial gait phase. Therefore, this embodiment provides high-quality observation input for subsequent Kalman filtering correction steps, enabling the lower limb exoskeleton robot to more accurately perceive the patient's real-time gait state. This provides a more reliable foundation for adaptive joint angle vector generation and compliant gait assistance, effectively improving the adaptability and gait bionics of the lower limb exoskeleton robot to complex environments.

[0089] In some preferred embodiments, the motion parameters are heel pressure or knee angle, and the formula for calculating the third initial gait phase corresponding to heel pressure is as follows:

[0090] ;

[0091] in, Indicates heel pressure. This indicates the preset coefficient corresponding to heel pressure. This indicates the third initial gait phase corresponding to heel pressure;

[0092] The formula for calculating the third initial gait phase corresponding to the knee joint angle is as follows:

[0093] ;

[0094] in, Indicates the knee joint angle. This represents the preset coefficient corresponding to the knee joint angle. This represents the third initial gait phase corresponding to the knee joint angle. This indicates the preset phase offset corresponding to the knee joint angle.

[0095] In this embodiment, heel pressure can be measured using a pressure sensor installed on the sole of the foot, and this heel pressure can reflect the start and end of the stance phase of the gait. Knee angle in this embodiment can be measured using a joint encoder installed at the knee joint, and the variation pattern of the knee angle is closely related to the swing and stance phases in the gait cycle.

[0096] In some preferred embodiments, the formula for calculating the desired joint angle vector is as follows:

[0097] ;

[0098] Where t represents time. Let z(t) represent the time scaling factor, and z(t) represent the auxiliary vector. The derivative of the auxiliary vector, and Both represent gain constants. This represents the desired joint angle vector. Let denote the derivative of the desired joint angle vector, and g denote the reference joint angle vector. Indicates the second gait phase. This represents the nonlinear forcing term corresponding to the second step phase;

[0099] The formula for calculating the nonlinear forcing term is shown below:

[0100] ;

[0101] in, Let N represent the Gaussian radial basis functions, and let N be the number of Gaussian radial basis functions. This represents the width of the i-th Gaussian radial basis function. This represents the center of the i-th Gaussian radial basis function. This represents the weight corresponding to the i-th Gaussian radial basis function.

[0102] The weight corresponding to the i-th Gaussian radial basis function in this embodiment is equivalent to the optimal weight in the above embodiment. That is, this embodiment is equivalent to obtaining the optimal weight of the adaptive joint angle vector generator by optimizing the weight corresponding to the Gaussian radial basis function using the taught gait data.

[0103] In some preferred embodiments, step S2 includes:

[0104] S21. Optimal weight reorganization of the adaptive joint angle vector generator based on all taught gait data using a local weighted regression algorithm.

[0105] The local weighted regression algorithm used in this embodiment is an existing algorithm, and its working principle and workflow will not be discussed in detail here.

[0106] As can be seen from the above, the joint angle vector generation method provided in this application enables the adaptive joint angle vector generator to have the ability to deeply learn and adaptively adjust to the gait characteristics of healthy people by pre-generating the optimal weight reorganization of the adaptive joint angle vector generator using the gait data of healthy people in different environments. It achieves phase prediction based on multi-sensor data by predicting the second gait phase based on the first gait data and multimodal motion data, thereby effectively improving the accuracy and reliability of phase estimation. Furthermore, it achieves the adjustment of the parameters of the adaptive joint angle vector generator based on the surrounding environment by first adjusting the reference joint velocity vector and time scaling factor of the adaptive joint angle vector generator according to visual data, and then using the adjusted adaptive joint angle vector generator to generate the desired joint angle based on the second gait phase, thereby dynamically adjusting the desired joint angle vector according to the surrounding environment. Therefore, this application is equivalent to providing a method that can generate joint angle vectors that are adapted to the current environment and have high gait biomimicry, thereby effectively meeting the rehabilitation needs of patients in real-life scenarios and avoiding the problems of increased gait abnormalities due to lack of environmental perception, which induce non-physiological movement patterns and affect the re-establishment of damaged neural pathways. This effectively improves the rehabilitation training effect and expands the application scope of rehabilitation training.

[0107] Secondly, this application also provides a gait control method for use in a lower limb exoskeleton robot, which includes the following steps:

[0108] B1. Acquire the first gait phase and multi-sensor data at the current moment; the first gait phase is the gait phase of the lower limb exoskeleton robot at the previous moment, and the multi-sensor data includes multimodal motion data and visual data;

[0109] B2. Predict the second gait phase based on the first gait phase and multimodal motion data; the second gait phase is the gait phase of the lower limb exoskeleton robot at the current moment;

[0110] B3. Adjust the reference joint angle vector and time scaling factor of the pre-optimized adaptive joint angle vector generator based on visual data, and then use the adjusted adaptive joint angle vector generator to generate the desired joint angle vector based on the second gait phase.

[0111] B4. Determine the current gait phase based on the second gait phase and multimodal motion data;

[0112] B5. Obtain the desired joint torque based on the desired joint angle vector;

[0113] B6. While keeping the desired joint torque constant, adjust the stiffness and damping of the lower limb exoskeleton robot according to the current gait stage;

[0114] The adaptive joint angle vector generator adopts a dynamic motion primitive architecture. The pre-optimization weight process of the adaptive joint angle vector generator includes:

[0115] S1. Collect teaching gait data of healthy individuals in different environments;

[0116] S2. Generate the optimal weight reassembly of the adaptive joint angle vector generator based on all taught gait data. The optimal weight reassembly includes the optimal weights corresponding to different lower limb joints.

[0117] This application provides a gait control method that combines a multi-sensor data fusion framework with an adaptive joint angle vector generator in a real-time environmental perception manner, and introduces a dynamic stiffness and damping adjustment mechanism based on gait phases, thereby achieving real-time perception of complex environments and compliant gait adaptation. Specifically, the fusion of multi-sensor data provides a comprehensive environmental perception foundation, solving the perception gap problem caused by relying solely on joint encoders in existing technologies; gait phase prediction combines historical phases and real-time motion information, improving the accuracy of estimation; the adaptive joint angle vector generator dynamically adjusts the reference joint angle vector and time scaling factor using visual data, and optimizes weights based on a dynamic motion primitive architecture and healthy human teaching data, ensuring that the generated desired joint angle vector has high biomimeticity and environmental adaptability; while the dynamic adjustment of stiffness and damping provides appropriate assistance according to the gait phase, providing high stiffness and low damping in the early stage of the support phase to ensure stability, and providing low stiffness and high damping in the swing phase to promote active patient participation. Through the above technical solutions, the lower limb exoskeleton robot can generate natural and adaptive gait in complex environments such as uphill and downhill scenarios. For example, when visual data detects an uphill section ahead, the system adjusts the reference joint angle vector and time scaling factor based on the visual information. This causes the adaptive joint angle vector generator to output the desired joint angle vector adapted to the uphill terrain. Simultaneously, the system dynamically adjusts stiffness and damping parameters according to the current gait phase. Specifically, in the initial stance phase, stiffness is increased and damping is decreased to provide strong support and prevent knee collapse; in the swing phase, stiffness is decreased and damping is increased to encourage the patient to actively swing their legs. Therefore, this application can effectively improve the naturalness and effectiveness of rehabilitation training, help better activate damaged neural pathways in patients, avoid the generation of abnormal gait patterns, and thus provide patients with a safer, more efficient, and personalized rehabilitation experience.

[0118] Specifically, the mapping relationship between the gait phase and the stiffness and damping of the lower limb exoskeleton robot in this embodiment is shown in the table below.

[0119] Mapping relationship between gait phase and stiffness and damping of lower limb exoskeleton robot

[0120]

[0121] In some preferred embodiments, step B6 includes:

[0122] B61. Obtain the torque for human-computer interaction;

[0123] B62. While keeping the desired joint torque constant, adjust the stiffness and damping of the lower limb exoskeleton robot according to the current gait stage and human-machine interaction torque.

[0124] The human-machine interaction torque in this embodiment refers to the torque of interaction between the lower limb exoskeleton robot and the patient's limbs. Its function is to quantify in real time the active or passive forces exerted by the patient on the robot during movement, thereby reflecting the patient's movement intentions and level of participation. This embodiment can obtain the human-machine interaction torque by installing force sensors or torque sensors on the lower limb exoskeleton robot to directly measure the torque at the joints. While keeping the desired joint torque constant, the system adjusts the stiffness and damping of the lower limb exoskeleton robot according to the current gait stage and the human-machine interaction torque to provide compliant gait assistance. This step is equivalent to dynamically adjusting the assistive characteristics of the lower limb exoskeleton robot based on the patient's real-time interaction state and gait stage to achieve more compliant and personalized gait assistance. Keeping the desired joint torque constant means that when adjusting stiffness and damping, the lower limb exoskeleton robot still targets the preset desired joint torque for control, but by changing the stiffness and damping parameters, its response characteristics to the desired joint torque and its compliance with the patient's active forces are affected. Stiffness and damping are key parameters describing the interaction between the exoskeleton and the patient. Stiffness reflects the exoskeleton's resistance to deviations from the desired position, while damping reflects its resistance to changes in the exoskeleton's motion speed. Adjusting these parameters can change the exoskeleton's "hardness" and how it follows or guides the patient's movements. For example, when an increase in the patient's active force is detected, the system can automatically reduce stiffness, thereby reducing exoskeleton intervention. This reduction can be achieved through a preset lookup table or a fuzzy logic-based controller.

[0125] The working principle of this embodiment is as follows: First, by acquiring the human-machine interaction torque, which directly reflects the active or passive force exerted by the patient on the lower limb exoskeleton robot at the current moment, the system quantifies the patient's movement intention and level of participation. Subsequently, the system combines the human-machine interaction torque with current gait stage information as the basis for adjusting the stiffness and damping of the lower limb exoskeleton robot. Throughout the adjustment process, it is expected that the joint torque remains constant, meaning that the exoskeleton system still strives to achieve the predetermined gait trajectory, but its implementation becomes more flexible and personalized. Specifically, when the system detects an increase in the patient's active force (human-machine interaction torque), indicating a strong movement intention, the system will correspondingly reduce the stiffness and increase the damping of the lower limb exoskeleton robot, thereby reducing the exoskeleton's intervention in the patient's movement, making the exoskeleton joints "softer," allowing the patient to lead the movement, and encouraging their active participation. Conversely, when the system detects that the patient is weak or exerts little force (the human-machine interaction torque is less than a preset torque threshold), it increases the stiffness and decreases the damping of the lower limb exoskeleton robot, making the exoskeleton joints "harder" and providing more auxiliary support to compensate for the patient's lack of strength. This embodiment, by introducing the human-machine interaction torque into the stiffness and damping adjustment strategy, overcomes the limitations of relying solely on gait phase adjustments. Therefore, this embodiment enables the lower limb exoskeleton robot's assistance to move beyond a rigid preset mode and dynamically respond based on the patient's real-time state. This adaptive adjustment mechanism, which combines gait phase information and the patient's active intentions, allows the lower limb exoskeleton robot to more accurately match the patient's individual needs, providing more compliant, personalized, and efficient gait assistance, thereby significantly improving the naturalness and effectiveness of rehabilitation training.

[0126] In some preferred embodiments, the formula for calculating the desired joint torque is as follows:

[0127] ;

[0128] in, Let J represent the desired joint torque, and J represent the Jacobian matrix of the lower limb skeletal robot. This represents the transpose of the Jacobian matrix of a lower limb skeletal robot. This represents the preset virtual stiffness. This represents the preset virtual damping. Let represent the desired joint angle vector, and q(t) represent the actual joint angle vector. This represents the derivative of the desired joint angle vector. The derivative of the actual joint angle vector (obtained by the joint encoder). This indicates the preset gravity compensation item.

[0129] In some preferred embodiments, step B4 includes:

[0130] B41. The gait phase corresponding to the preset phase mapping interval to which the second gait phase belongs and the preset conditions satisfied by the multimodal motion data is taken as the current gait phase.

[0131] The preset phase mapping interval refers to the range in which a continuous second gait phase is divided into several discrete gait sub-phases with specific physiological significance. The preset conditions refer to the logical rules or thresholds set based on multimodal motion data to assist or correct gait phase judgment. In this embodiment, the gait phase corresponding to the preset phase mapping interval to which the second gait phase belongs and the preset conditions satisfied by the multimodal motion data is taken as the current gait phase. For example, when the second gait phase ∈ [11π / 6, 2π) and the angular velocity of the inertial sensor set at the lower leg crosses zero, the end of the swing phase is taken as the current gait phase.

[0132] As can be seen from the above, the joint angle vector generation method and gait control method provided in this application, by pre-generating the optimal weight reorganization of the adaptive joint angle vector generator using the taught gait data of healthy individuals in different environments, enables the adaptive joint angle vector generator to possess the ability to deeply learn and adaptively adjust to the gait characteristics of healthy individuals. Phase prediction based on multi-sensor data is achieved by predicting the second gait phase according to the first gait data and multimodal motion data, thereby effectively improving the accuracy and reliability of phase estimation. Furthermore, by first adjusting the reference joint velocity vector and time scaling factor of the adaptive joint angle vector generator based on visual data, the root... The parameters of the adaptive joint angle generator are adjusted according to the surrounding environment. Then, the adjusted adaptive joint angle vector generator generates the desired joint angle based on the second gait phase. This achieves dynamic adjustment of the desired joint angle vector according to the surrounding environment. Therefore, this application provides a method to generate joint angle vectors that are adapted to the current environment and have high gait biomimicry. This effectively meets the rehabilitation needs of patients in real-life scenarios and avoids the problem of increased gait abnormalities due to lack of environmental awareness, which can induce non-physiological movement patterns and affect the re-establishment of damaged neural pathways. This effectively improves the rehabilitation training effect and expands the application scope of rehabilitation training.

[0133] In the embodiments provided in this application, it should be understood that relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0134] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A joint angle vector generation method applied in a lower extremity exoskeleton robot, characterized in that, The joint angle vector generation method comprises the following steps: A1, acquiring a first gait phase and multi-sensor data at a current time; the first gait phase is a gait phase of the lower extremity exoskeleton robot at a previous time, and the multi-sensor data comprises multi-modal motion data and visual data; A2, predicting a second gait phase according to the first gait phase and the multi-modal motion data; the second gait phase is a gait phase of the lower extremity exoskeleton robot at the current time; A3, adjusting a reference joint angle vector and a time scaling factor of a pre-optimized adaptive joint angle vector generator according to the visual data, and then generating an expected joint angle vector according to the second gait phase by using the adjusted adaptive joint angle vector generator; The adaptive joint angle vector generator adopts a dynamic motion primitive architecture, and a pre-optimized weight process of the adaptive joint angle vector generator comprises: S1, collecting demonstration gait data of healthy people in different environments; S2, generating an optimal weight set of the adaptive joint angle vector generator according to all the demonstration gait data, wherein the optimal weight set comprises optimal weights corresponding to different lower limb joints.

2. The articular angle vector generation method according to claim 1, characterized by, Step A2 comprises: A21, generating a first initial gait phase according to the first gait phase by using a state transition equation of an extended Kalman filter; the first initial gait phase is a predicted value of the gait phase at the current time generated according to the first gait phase; A22, generating a second initial gait phase according to the multi-modal motion data by using an observation equation of the extended Kalman filter; the second initial gait phase is an observed value of the gait phase at the current time generated according to the multi-modal motion data; A23, correcting the first initial gait phase according to the second initial gait phase by using the extended Kalman filter to obtain the second gait phase.

3. The articular angle vector generation method according to claim 2, characterized by, The multi-modal motion data comprises a plurality of motion parameters, and step A22 comprises: A221, for each motion parameter, generating a third initial gait phase corresponding to the motion parameter according to the motion parameter by using an observation equation of the extended Kalman filter; A222, acquiring the second initial gait phase according to all the third initial gait phases.

4. The articular angle vector generation method according to claim 3, characterized by, The motion parameter is a heel pressure or a knee joint angle, and a calculation formula of the third initial gait phase corresponding to the heel pressure is as follows: ； wherein, represents a heel pressure, represents a preset coefficient corresponding to the heel pressure, represents a third initial gait phase corresponding to the heel pressure; A calculation formula of the third initial gait phase corresponding to the knee joint angle is as follows: ； wherein, represents a knee joint angle, represents a preset coefficient corresponding to the knee joint angle, represents a third initial gait phase corresponding to the knee joint angle, represents a preset phase offset corresponding to the knee joint angle.

5. The articular angle vector generation method of claim 1, wherein, A calculation formula of the expected joint angle vector is as follows: ； where t denotes time, denotes a time scaling factor, z(t) denotes a helper vector, denotes a derivative of the helper vector, and both denote a gain constant, denotes a desired joint angle vector, denotes a derivative of the desired joint angle vector, g denotes a reference joint angle vector, denotes a second gait phase, denotes a nonlinear forcing term corresponding to the second gait phase; A calculation formula of the non-linear forcing term is as follows: ； wherein, represents a Gaussian radial basis function, N represents a number of Gaussian radial basis functions, represents a width of the i-th Gaussian radial basis function, represents a center of the i-th Gaussian radial basis function, represents a weight corresponding to the i-th Gaussian radial basis function.

6. The articular angle vector generation method of claim 1, wherein, Step S2 comprises: S21, generating the optimal weight set of the adaptive joint angle vector generator according to all the demonstration gait data by using a local weighted regression algorithm.

7. A gait control method characterized by, The gait control method is applied in a lower extremity exoskeleton robot, and the gait control method comprises the following steps: B1, acquiring a first gait phase and multi-sensor data at a current time; the first gait phase is a gait phase of the lower extremity exoskeleton robot at a previous time, and the multi-sensor data comprises multi-modal motion data and visual data; B2, predicting a second gait phase according to the first gait phase and the multi-modal motion data; the second gait phase is a gait phase of the lower extremity exoskeleton robot at a current time; B3, adjusting a reference joint angle vector and a time scaling factor of a pre-optimized adaptive joint angle vector generator according to the visual data, and then generating a desired joint angle vector according to the second gait phase by using the adjusted adaptive joint angle vector generator; B4, determining a current gait stage according to the second gait phase and the multi-modal motion data; B5, obtaining a desired joint torque according to the desired joint angle vector; B6, adjusting a stiffness and a damping of the lower extremity exoskeleton robot according to the current gait stage while keeping the desired joint torque unchanged; The adaptive joint angle vector generator adopts a dynamic motion primitive architecture, and a pre-optimized weight process of the adaptive joint angle vector generator includes: S1, collecting teaching gait data of healthy people in different environments; S2, generating an optimal weight set of the adaptive joint angle vector generator according to all the teaching gait data, the optimal weight set including optimal weights corresponding to different lower extremity joints.

8. The gait-controlling method according to claim 7, characterized by, Step B6 includes: B61, obtaining a human-robot interaction torque; B62, adjusting the stiffness and the damping of the lower extremity exoskeleton robot according to the current gait stage and the human-robot interaction torque while keeping the desired joint torque unchanged.

9. The gait-controlling method according to claim 7, characterized by, The calculation formula of the desired joint torque is as follows: ； wherein, denotes a desired joint torque, J denotes a Jacobian matrix of the lower-limb skeletal robot, denotes a transpose of the Jacobian matrix of the lower-limb skeletal robot, denotes a preset virtual stiffness, denotes a preset virtual damping, denotes a desired joint angle vector, q(t) denotes an actual joint angle vector, denotes a derivative of the desired joint angle vector, denotes a derivative of the actual joint angle vector, denotes a preset gravity compensation term.

10. The gait-controlling method according to claim 7, wherein Step B4 includes: B41, taking a gait stage corresponding to a preset phase mapping interval to which the second gait phase belongs and a preset condition satisfied by the multi-modal motion data as a current gait stage.

Images