CN-121989255-A - Control method, device, equipment and storage medium for lower limb exoskeleton robot

CN121989255ACN 121989255 ACN121989255 ACN 121989255ACN-121989255-A

Abstract

The application belongs to the technical field of robot control and discloses a control method, a device, equipment and a storage medium of a lower limb exoskeleton robot, wherein the method comprises the steps of constructing and obtaining topographic feature information corresponding to surrounding topographic images according to an internal reference matrix of a camera and a coordinate system conversion matrix, identifying and obtaining target motion intention of a target user based on a Gaussian mixture model in combination with surface electromyographic signals and kinematic data, carrying out motion track prediction based on the topographic feature information, the target motion intention and body condition parameters of the target user to obtain expected joint motion track information of the target user, carrying out moment compensation calculation according to the kinematic data and dynamic compensation factors by adopting a model-free prediction control method to obtain target control moment when the lower limb exoskeleton robot moves along the expected joint motion track information so as to control the lower limb exoskeleton robot; by the method, the control precision and efficiency of the lower limb exoskeleton robot are improved.

Inventors

YANG PENG
DENG TAO
WANG HAO
GU JIAWEI
XU HUJUN

Assignees

季华实验室

Dates

Publication Date: 20260508
Application Date: 20260401

Claims (10)

1. The control method of the lower limb exoskeleton robot is characterized by comprising the following steps: Acquiring surface electromyographic signals and surrounding topographic images of a target user, wherein the target user is a user of the lower limb exoskeleton robot; According to the pixel information in the surrounding topographic image, combining preset camera internal reference matrix information and preset coordinate system conversion matrix information, and carrying out feature extraction to obtain topographic feature information corresponding to the surrounding topographic image; based on a preset Gaussian mixture model, combining the surface electromyographic signals and the kinematic data to perform intention recognition so as to obtain the target movement intention of the target user; Track prediction is carried out based on the topographic feature information, the target movement intention and the physical condition parameters of the target user, so that expected joint movement track information of the target user is obtained; Performing moment compensation calculation by adopting a model-free predictive control method according to the kinematic data and combining dynamic compensation factors to obtain a target control moment when the lower limb exoskeleton robot moves along the information of the movement track of the expected joint; And controlling the lower limb exoskeleton robot according to the target control moment.
2. The method according to claim 1, wherein the feature extraction is performed according to pixel information in the surrounding topographic image in combination with preset camera internal reference matrix information and preset coordinate system conversion matrix information to obtain topographic feature information corresponding to the surrounding topographic image, comprising: converting pixel information in the surrounding terrain image into three-dimensional coordinates of a coordinate system corresponding to the lower limb exoskeleton robot by using preset camera internal reference matrix information and preset coordinate system conversion matrix information; Constructing a terrain elevation map centered on the lower extremity exoskeleton robot based on the three-dimensional coordinates; And extracting key features based on the continuously obtained multi-frame topographic elevation map to obtain topographic feature information corresponding to the surrounding topographic images.
3. The method according to claim 1, wherein performing intention recognition based on a preset gaussian mixture model in combination with the surface electromyographic signals and the kinematic data to obtain a target movement intention of the target user comprises: constructing and obtaining a surface myoelectricity characteristic vector corresponding to the surface myoelectricity signal; Extracting angular velocity data and acceleration data from the kinematic data; Fusing the angular velocity data and the acceleration data to obtain a kinematic feature vector corresponding to the kinematic data; and inputting the surface myoelectricity feature vector and the kinematic feature vector into a preset Gaussian mixture model for intention recognition to obtain the target movement intention of the target user.
4. The method according to claim 3, wherein the step of inputting the surface myoelectricity feature vector and the kinematic feature vector to a preset gaussian mixture model for intention recognition to obtain the target movement intention of the target user comprises: Inputting the surface myoelectricity feature vector and the kinematic feature vector into a preset Gaussian mixture model for calculation to obtain Gaussian distribution probabilities of various motion intentions corresponding to the target user; Calculating by adopting a preset posterior probability calculation method according to the Gaussian distribution probability to obtain posterior probabilities of various movement intentions corresponding to the target user; and determining the motion intention corresponding to the maximum value in the posterior probability as the target motion intention of the target user.
5. The method according to claim 1, wherein performing trajectory prediction based on the topographic feature information, the target motion intent, and the physical condition parameters of the target user to obtain desired joint motion trajectory information of the target user, comprises: acquiring physical condition parameters of the target user; fusing the topographic feature information, the target movement intention and the physical condition parameters of the target user to obtain a fusion condition vector; according to the plantar contact state of the lower limb exoskeleton robot in the kinematic data, carrying out phase prediction to obtain a predicted gait phase of the lower limb exoskeleton robot; And predicting the motion trail based on the fusion condition vector and the predicted gait phase to obtain the expected joint motion trail information of the target user.
6. The method according to claim 1, wherein the step of performing a moment compensation calculation by combining a dynamic compensation factor according to the kinematic data by using a model-free predictive control method to obtain a target control moment when the lower limb exoskeleton robot moves along the desired joint movement track information comprises: Based on the historical joint angle difference value and the historical control increment of the lower limb exoskeleton robot at adjacent moments in the kinematic data, carrying out iterative updating on the pseudo jacobian matrix of the lower limb exoskeleton robot by combining dynamic compensation factors to obtain an optimized pseudo jacobian matrix; Calculating by adopting a model-free predictive control method based on the optimized pseudo jacobian matrix and combining with a preset optimal control moment increment calculation model to obtain an optimal control moment increment when the lower limb exoskeleton robot moves along the expected joint movement track information; Determining and obtaining initial control moment when the lower limb exoskeleton robot moves along the expected joint movement track information; and calculating based on the optimal control moment increment and the initial control moment to obtain a target control moment when the lower limb exoskeleton robot moves along the information of the expected joint movement track.
7. The method according to claim 6, wherein the calculating, by using a model-free predictive control method, based on the optimized pseudo jacobian matrix and in combination with a preset optimal control moment increment calculation model, obtains an optimal control moment increment when the lower limb exoskeleton robot moves along the desired joint movement track information, includes: based on the optimized pseudo-Jacobian matrix, calculating by combining with an actual joint angle vector of the lower limb exoskeleton robot to obtain a joint angle prediction error when the lower limb exoskeleton robot moves along the expected joint movement track information; And inputting the joint angle prediction error into a preset optimal control moment increment calculation model for calculation to obtain the optimal control moment increment when the lower limb exoskeleton robot moves along the information of the expected joint movement track.
8. A lower extremity exoskeleton robot control device, comprising: The system comprises an acquisition module, a control module and a control module, wherein the acquisition module is used for acquiring surface electromyographic signals and surrounding topographic images of a target user and kinematic data of a lower limb exoskeleton robot; the extraction module is used for carrying out feature extraction according to pixel information in the surrounding topographic image and combining preset camera internal reference matrix information and preset coordinate system conversion matrix information to obtain topographic feature information corresponding to the surrounding topographic image; The recognition module is used for carrying out intention recognition based on a preset Gaussian mixture model and combining the surface electromyographic signals and the kinematic data to obtain the target movement intention of the target user; the prediction module is used for carrying out track prediction based on the topographic feature information, the target movement intention and the physical condition parameters of the target user to obtain expected joint movement track information of the target user; The calculation module is used for carrying out moment compensation calculation according to the kinematic data and the dynamic compensation factors by adopting a model-free predictive control method to obtain a target control moment when the lower limb exoskeleton robot moves along the information of the movement track of the expected joint; And the control module is used for controlling the lower limb exoskeleton robot according to the target control moment by the calculation module.
9. An electronic device comprising a processor and a memory, the memory storing a computer program executable by the processor, when executing the computer program, running the steps in the method of controlling a lower extremity exoskeleton robot as claimed in any one of claims 1 to 7.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, runs the steps in the lower extremity exoskeleton robot control method of any one of claims 1 to 7.

Description

Control method, device, equipment and storage medium for lower limb exoskeleton robot Technical Field The application relates to the technical field of robot control, in particular to a method, a device, equipment and a storage medium for controlling a lower limb exoskeleton robot. Background The lower limb exoskeleton robot is used as wearable electromechanical integrated equipment, aims to provide assistance through coupling with lower limbs of a human body, and has the core challenge of realizing safe, natural and efficient man-machine cooperative motion. Currently mainstream control methods, such as tracking control based on preprogrammed trajectories, phase control based on finite state machines, and force/bit mixture control based on models, all expose significant limitations in practical applications. Specifically, although the tracking control method based on the preprogrammed track is stable in performance in a fixed and structured environment, the preset joint angle track cannot adapt to environmental changes or changes of user intention, so that the gait strategy cannot be actively adjusted when the system faces complex terrain or sudden actions of the user, and the adaptability is extremely poor. The finite state machine based phase control method, while effective for periodic gait, has difficulty in handling aperiodic, discrete motion mode switching, such as transitioning from walking on level ground to ascending or descending stairs, which makes it difficult for an exoskeleton robot to achieve smooth and natural motion in complex and diverse environments. The effectiveness of the model-based force/position hybrid control approach relies heavily on accurate human and exoskeleton dynamics models. However, accurate kinetic models are difficult to obtain and model parameters can change over time with individual user differences, load changes, and terrain changes, resulting in degraded or even unstable control performance. Therefore, the problems of environmental perception deficiency, man-machine interaction stiffness and strong model dependence commonly exist in the prior art, so that intelligent, flexible and efficient power assisting of the lower limb exoskeleton robot in a real complex scene is difficult to realize. In view of the above, there is a need in the art for improvements. Disclosure of Invention The application aims to provide a lower limb exoskeleton robot control method, device, equipment and storage medium, wherein expected joint movement track information is obtained through topographic feature information, target movement intention and physical condition parameter prediction of a target user, then a model-free prediction control method is adopted, and a target control moment when the lower limb exoskeleton robot moves along the expected joint movement track information is calculated according to kinematic data and dynamic compensation factors, so that the lower limb exoskeleton robot is controlled, the problem that the lower limb exoskeleton robot is difficult to accurately control due to the defects of environmental perception deficiency, human-computer interaction stiffness and strong model dependence existing in the existing lower limb exoskeleton robot control is solved, the lower limb exoskeleton robot can be accurately controlled to move along the expected joint movement track information by the model-free prediction control method, and the control precision and efficiency of the lower limb exoskeleton robot are improved. In a first aspect, the present application provides a method for controlling a lower extremity exoskeleton robot, comprising: Acquiring surface electromyographic signals and surrounding topographic images of a target user, wherein the target user is a user of the lower limb exoskeleton robot; According to the pixel information in the surrounding topographic image, combining preset camera internal reference matrix information and preset coordinate system conversion matrix information, and carrying out feature extraction to obtain topographic feature information corresponding to the surrounding topographic image; based on a preset Gaussian mixture model, combining the surface electromyographic signals and the kinematic data to perform intention recognition so as to obtain the target movement intention of the target user; Track prediction is carried out based on the topographic feature information, the target movement intention and the physical condition parameters of the target user, so that expected joint movement track information of the target user is obtained; Performing moment compensation calculation by adopting a model-free predictive control method according to the kinematic data and combining dynamic compensation factors to obtain a target control moment when the lower limb exoskeleton robot moves along the information of the movement track of the expected joint; And controlling the lower limb exoskeleton robot according to the target control momen