CN-121987458-A - Lower limb gait generation method and device based on dynamic motion primitive

CN121987458ACN 121987458 ACN121987458 ACN 121987458ACN-121987458-A

Abstract

The application provides a lower limb gait generating method and device based on dynamic motion primitives, wherein the method comprises the steps of determining model input parameters based on target terrain parameters and target human motion parameters, obtaining a first direction position curve of target lower limb tail end motion based on the model input parameters and a first dynamic motion primitive model of a target terrain type, obtaining a second direction position curve of target lower limb tail end motion based on the model input parameters and a second dynamic motion primitive model of the target terrain type, and generating the target lower limb gait based on the first direction position curve and the second direction position curve of the target lower limb tail end motion. The method of the application is based on a dynamic motion primitive model, rapidly generates personalized, continuous and natural lower limb gait curves adapting to different topographic parameters and pace speeds, accords with the motion rule of human bodies, and can provide accurate track planning input for exoskeleton, intelligent artificial limbs and other devices.

Inventors

WANG XINGJIAN
Chen Nuobi
ZHANG YIXIN
ZHANG YUWEI
WANG SHAOPING
WANG JIAQI

Assignees

天目山实验室

Dates

Publication Date: 20260508
Application Date: 20251222

Claims (10)

1. The lower limb gait generating method based on the dynamic motion primitive is characterized by comprising the following steps of: Determining model input parameters based on the target topographic parameters and the target human body motion parameters; Obtaining a first direction position curve of the movement of the tail end of the target lower limb based on the model input parameters and a first dynamic movement primitive model of the target terrain type generated in advance; Obtaining a second direction position curve of the movement of the tail end of the target lower limb based on the model input parameters and a second dynamic movement primitive model of the target terrain type, which is generated in advance; and generating the gait of the target lower limb based on the first direction position curve and the second direction position curve of the target lower limb tail end movement.
2. The method according to claim 1, wherein the method further comprises: collecting human body lower limb movement tracks under a plurality of terrain types, and processing the human body lower limb movement tracks to obtain a two-dimensional lower limb tail end movement track, wherein the terrain types comprise a flat land, an ascending ramp, a descending ramp, a stair ascending channel and a stair descending channel; generating a first dynamic motion primitive model of each terrain type by using a first direction track of a two-dimensional lower limb tail end motion track; and generating a second dynamic motion primitive model of each terrain type by using a second direction track of the two-dimensional lower limb tail end motion track.
3. The method of claim 2, wherein the steps of acquiring the motion trajectories of the lower limbs of the human body under a plurality of terrain types, processing the motion trajectories of the lower limbs of the human body to obtain the motion trajectories of the tail ends of the lower limbs, and the method comprises the steps of: Setting a plurality of reflective marker points on the lower limb part of the subject; acquiring three-dimensional trajectories of a plurality of reflective marker points when a subject walks in the terrain of each terrain type by using a Vicon system, so as to obtain a human lower limb movement trajectory; processing the motion trail of the lower limb of the human body to obtain a two-dimensional motion trail of the tail end of the lower limb: , Is the step length; The first direction track of the two-dimensional lower limb tail end motion track is as follows: The second direction track of the two-dimensional lower limb tail end motion track is as follows: 。
4. The method of claim 3, wherein generating a first dynamic motion primitive model for each terrain type using a first directional trajectory of a two-dimensional extremity motion trajectory comprises: Construction No. Gaussian base function : Wherein, the And Is that Center point and width of (a); Is a standard phase variable; ; the number of Gaussian basis functions; Based on the first Gaussian base function Weights of (2) Is a square loss function of (2), determines the weight ; Calculating a first kernel function : By applying a first kernel function Mapping onto time domain to obtain a first core function , Is a time variable; Based on a first core function Generating a first dynamic motion primitive model: Wherein, the Is the initial position of the first direction track, Is the termination position of the first direction track; Is a time constant; and (3) with Respectively an elastic coefficient and a damping coefficient; Is a first directional position curve; is a first directional velocity profile.
5. The method according to claim 4, wherein the method further comprises: Solving differential equations , To adjust the factor, a mapping function is determined ; Dividing the time interval [0, T ] into N equally spaced time sequences Using mapping functions Obtaining the corresponding standard phase variable sequence as ; Calculating the center point of the ith Gaussian basis function And width of : , 。
6. The method of claim 5, wherein the method is based on an ith Gaussian basis function Weights of (2) Is a square loss function of (2), determines the weight The method comprises the following steps: Constructing the ith Gaussian basis function Weights of (2) Is the square loss function of (2) : Wherein, the Representing a total number of time steps; ; Function value The calculation formula of (2) is as follows: For the ith Gaussian basis function After conversion to the time domain A value of (a); The calculation formula of (2) is as follows: solving a square loss function Is the minimum of (2) to obtain : Wherein the vector is The method comprises the following steps: Matrix array The method comprises the following steps: vector quantity The method comprises the following steps: 。
7. the method of claim 4, wherein generating a second dynamic motion primitive model for each terrain type using a second directional trajectory of the two-dimensional extremity motion trajectory comprises: Calculating a second kernel function by using a plurality of second direction tracks of the two-dimensional lower limb tail end motion tracks , Is a standard phase variable; The second core function Mapping onto time domain to obtain a second core function , Is a time variable; Based on a second core function Generating a second dynamic motion primitive model: Wherein, the Is the initial position of the second direction track, Is the termination position of the second direction track; Is a time constant; and (3) with Respectively an elastic coefficient and a damping coefficient; Is a second directional position curve; is a second directional velocity profile.
8. The method of claim 7, wherein the target terrain parameter is a grade or a stair step height; determining model input parameters based on the target terrain parameters and the target human motion parameters, comprising: determining a termination position of a first directional trajectory of a first dynamic motion primitive model based on a target terrain parameter And an ending position of a second directional trajectory of the second dynamic motion primitive model ; Determining a first dynamic motion primitive model and a time constant in the first dynamic motion primitive model according to the target human motion parameters 。
9. The method of claim 1, wherein generating the target lower limb gait based on the first and second directional position curves of the target lower limb extremity movement comprises: obtaining the length of the thigh of the lower limb of the human body in the sagittal plane And leg length ; Obtaining a first direction position curve and a second direction position curve based on the movement of the tail end of the target lower limb Time of day position of lower extremity ; Calculating ankle angle : Calculating hip joint angle : Calculating knee joint angles : Based on hip joint angle And knee joint angle A target lower limb gait is generated.
10. A lower limb gait generating device based on dynamic motion primitives, comprising: the determining unit is used for determining model input parameters based on the target terrain parameters and the target human motion parameters; the first processing unit is used for obtaining a first direction position curve of the movement of the tail end of the target lower limb based on the model input parameters and a pre-generated first dynamic movement primitive model of the target terrain type; The second processing unit is used for obtaining a second direction position curve of the movement of the tail end of the target lower limb based on the model input parameters and a pre-generated second dynamic movement primitive model of the target terrain type; and the gait generating unit is used for generating the target lower limb gait based on the first direction position curve and the second direction position curve of the target lower limb tail end movement.

Description

Lower limb gait generation method and device based on dynamic motion primitive Technical Field The application relates to the technical field of human motion analysis, in particular to a lower limb gait generating method and device based on dynamic motion elements. Background The core goal of wearable walker devices (e.g., exoskeletons, intelligent prostheses) is to provide natural, coordinated motion assistance to the user. The key to achieving this is that the device is able to generate a gait trajectory that matches the user's current movement intent and environment. The conventional gait generation method mainly comprises the following steps of firstly, a fixed track library based on pre-programming, which lacks adaptability and cannot cope with complex and changeable terrains and individual differences, secondly, generating the gait by solving a multi-body dynamics equation based on a fine biomechanical model, wherein the model is complex, has a plurality of parameters and large calculation amount, and is difficult to realize real-time control, thirdly, a black box model based on machine learning, although the gait characteristics can be learned from data, the interpretability is poor, and the generated track has an ambiguous physical meaning, and the smoothness and the stability are sometimes difficult to ensure due to the fact that a large amount of labeled data are relied on. Disclosure of Invention In view of the above, the present application provides a lower limb gait generating method and device based on dynamic motion primitives, so as to solve the above technical problems. In a first aspect, an embodiment of the present application provides a method for generating gait of a lower limb based on dynamic motion primitives, including: Determining model input parameters based on the target topographic parameters and the target human body motion parameters; Obtaining a first direction position curve of the movement of the tail end of the target lower limb based on the model input parameters and a first dynamic movement primitive model of the target terrain type generated in advance; Obtaining a second direction position curve of the movement of the tail end of the target lower limb based on the model input parameters and a second dynamic movement primitive model of the target terrain type, which is generated in advance; and generating the gait of the target lower limb based on the first direction position curve and the second direction position curve of the target lower limb tail end movement. In one possible implementation, the method further includes: collecting human body lower limb movement tracks under a plurality of terrain types, and processing the human body lower limb movement tracks to obtain a two-dimensional lower limb tail end movement track, wherein the terrain types comprise a flat land, an ascending ramp, a descending ramp, a stair ascending channel and a stair descending channel; generating a first dynamic motion primitive model of each terrain type by using a first direction track of a two-dimensional lower limb tail end motion track; and generating a second dynamic motion primitive model of each terrain type by using a second direction track of the two-dimensional lower limb tail end motion track. In one possible implementation, the method for acquiring the motion trajectories of the lower limbs of the human body under a plurality of terrain types, processing the motion trajectories of the lower limbs of the human body to obtain the motion trajectories of the tail ends of the lower limbs comprises the following steps: Setting a plurality of reflective marker points on the lower limb part of the subject; acquiring three-dimensional trajectories of a plurality of reflective marker points when a subject walks in the terrain of each terrain type by using a Vicon system, so as to obtain a human lower limb movement trajectory; processing the motion trail of the lower limb of the human body to obtain a two-dimensional motion trail of the tail end of the lower limb: , Is the step length; The first direction track of the two-dimensional lower limb tail end motion track is as follows: The second direction track of the two-dimensional lower limb tail end motion track is as follows: 。 In one possible implementation, the generating a first dynamic motion primitive model for each terrain type using a first directional trajectory of a two-dimensional lower extremity motion trajectory includes: Construction No. Gaussian base function: Wherein, the AndIs thatCenter point and width of (a); Is a standard phase variable; ; the number of Gaussian basis functions; Based on the first Gaussian base functionWeights of (2)Is a square loss function of (2), determines the weight; Calculating a first kernel function: By applying a first kernel functionMapping onto time domain to obtain a first core function,Is a time variable; Based on a first core function Generating a first dynamic motion primitive model