Disclosure of Invention
Aiming at the problem that gait of the lower limb exoskeleton robot is assisted as required, the invention provides a multi-mode hybrid control method based on the lower limb exoskeleton robot, firstly, a representation method of a track error in a passive mode and a contour error in an active mode is provided according to the characteristics of different control modes, and then a controller with dynamic compensation is designed based on a dynamic model of the lower limb exoskeleton robot, so that motion control in different control modes is realized.
The auxiliary control method based on the lower limb exoskeleton robot as required comprises the following specific steps:
step one, aiming at a patient, taking a motion track of an ankle joint of the patient in a three-dimensional space as an expected motion track;
I.e.
Where s.epsilon.0,100 represents the percentage of the current motion time T with respect to the gait cycle T,Representing a three-dimensional euclidean space.
And step two, measuring the movement angle of the robot by utilizing an angle sensor on the robot, and obtaining the actual position Pa (t) of the ankle joint of the robot through kinematic calculation.
And step three, under the passive mode, the robot drives the limb of the patient to move, and the expected movement track of the patient is used for calculating a track tracking error ep1.
The track tracking error ep1 is expressed as:
ep1＝Pa-f(s)
Step four, in the active mode, calculating the distance from the nearest point f (s*) to the current actual position Pa (t), namely the contour tracking error ep2, on the expected motion track f(s) of the patient;
I.e.
ep2＝Pa-f(s*)
The nearest point f (s*) to the current actual position is calculated using the following controller:
k and lambda are positive constants, kΨ is a function of s, sigma is a sliding mode surface function, alpha is the order, alpha is equal to or greater than 1, and ψ is a variable describing distance projection.
And fifthly, designing a controller with a dynamics model and speed error estimation by utilizing a track tracking error ep1 and a contour tracking error ep2 based on a dynamics model of the lower limb exoskeleton robot, and realizing motion control under different control modes.
The kinetic equation of the robot is:
Wherein M is an inertia matrix, C is a Golgi force and centripetal force term, G is an attractive force term, F is a friction force, τext is an interaction force between the robots, namely force applied to the robots by the patients, q= [ thetah θk]T ] is a generalized variable, wherein thetah and thetak are angles of hip joints and knee joints respectively, τ is a control law and is designed as follows:
J is jacobian of the robot, Kd is the speed gain,As an estimate of the speed error,As an estimated value of the dynamics model, Fa is a moment term, and in the passive control mode:
Fa＝-Kpep
Kp is the position gain, ep is the error, and at this point ep＝ep1;
F in active control modea＝k1ω1Fac+k2ω2Ftr
K1 and k2 are control gains for adjusting the magnitude of the output torque, ω1 and ω2 are the weights of the tangential component and the normal component, respectively, r is a variable for adjusting the relative weights of the two components according to the attitude profile error ep;
Fac and Ftr are unit vectors for applying an adjusting moment direction at the nearest point, and the calculation formula is as follows:
Fac＝-(n+b)/||n+b||＝-ep/||ep||
Ftr=t
n and b represent the normal vector and the secondary normal vector, respectively, at the nearest position, ep is the error, at which point ep＝ep2, t is the tangent vector at the nearest position.
The invention has the advantages that:
1) The multi-mode hybrid control method based on the lower limb exoskeleton robot is characterized in that a passive controller based on tracking errors and an active controller based on contour errors are designed to meet the training requirements of different rehabilitation stages, and the two controllers adopt control architectures with identical structures and only have different error expression modes, so that different errors can be input based on different rehabilitation training requirements, and the switching of training modes is realized.
Detailed Description
The invention is further illustrated in the following figures and examples.
The auxiliary control method based on the lower limb exoskeleton robot as required is shown in fig. 1, and comprises the following specific steps:
step one, aiming at a patient, taking a motion track of a lower limb tail end point, namely an ankle joint, of the patient in a three-dimensional space as an expected motion track;
Namely:
wherein s ε [0,100] represents the percentage of current motion time T relative to gait cycle T; Representing a three-dimensional euclidean space.
And step two, measuring the movement angle of the robot by utilizing an angle sensor on the robot, and obtaining the actual end point of the robot, namely the actual position Pa (t) of the ankle joint point through kinematic calculation.
And step three, under the passive mode, the robot drives the limb of the patient to move, and the expected movement track of the patient is used for calculating a track tracking error ep1.
At this time, the control is track tracking error, and the expected motion point is
Pd1(t)=f(s) (2)
At this time s is
Is a quantity related to the current run time t. The track tracking error is expressed as
The track tracking error ep1 is expressed as:
ep1＝Pa-Pd1＝Pa-f(s) (4)
pd1 (t) is an expected movement point when the robot drives the limbs of the patient to move;
Step four, in the active mode, calculating the distance from the nearest point f (s*) to the current actual position Pa (t), namely the contour tracking error ep2, on the expected motion track f(s) of the patient;
I.e.
ep2＝Pa-Pd2＝Pa-f(s*) (5)
Pd2 is the closest point on the desired motion trajectory to the current actual position, the distance of this point to the actual position Pa (t);
In order to solve the nearest point f (s*) on the desired motion trajectory f(s), the following method is adopted:
If mappingDefining a three-dimensional Euclidean spaceA parameterized curve C for the actual positionAnd is also provided withThe controller calculates as follows:
The nearest location point f (s*) to the actual location on the whole curve C can be found as shown in fig. 2, where k and λ are positive constants, kΨ is a function of s, Σ is a sliding mode surface function, and will be in the form of the proof process as follows:
for Frenet frame at s ε [0,l ] is { f(s); t(s), n(s), b(s) } satisfies
Wherein the method comprises the steps of
t(s)=fs(s)/||fs(s)|| (8)
Definition:
from the definition, ψ describes the projection length of Γ at fs, when satisfied at the nearest point s*
At the same time get by definition
Can be obtained by using (7) and (11)
Definition of the definitionThenDefine an approach law as
Substituting (6) to obtain
Selecting Lyapunov function as
Thereby making it
The calculation flow of the algorithm is shown in fig. 3.
And fifthly, designing a controller with a dynamics model and speed error estimation by utilizing a track tracking error ep1 and a contour tracking error ep2 based on a dynamics model of the lower limb exoskeleton robot, and realizing motion control under different control modes.
The kinetic equation of the robot is:
Wherein M is an inertia matrix, C is a Golgi force and centripetal force term, G is an attractive force term, F is a friction force, τext is an interaction force between the robots, namely the force applied to the robots by the patients, q= [ thetah θk]T ] is a generalized variable, wherein thetah and thetak are angles of hip joints and knee joints respectively, and τ is a control law;
The requirements are as follows:
Property 1:M is a positive symmetry matrix;
Properties 2:M and C satisfy:
Definition e=q-qd, henceI.e. the two satisfy the jacobian relationship, then
Wherein the method comprises the steps ofThe dynamic model and friction are difficult to model accurately.
In a practical system, for the first derivative eitherOr is alsoAre difficult to directly measure and thenIs also difficult to solve, butIs bounded and exists inThus using a first order filter for estimation, i.e
Introducing a measurable auxiliary signal s
Thereby making it
The control law is designed as
J is jacobian of the robot, Kd is the speed gain,As an estimate of the speed error,Is an estimated value of the dynamics model;
Wherein the method comprises the steps ofAs an estimate of D,The estimation error is bounded, existsAdopting RBF neural network approximation processing to obtain
Definition of the definitionFor the estimated value, the estimated optimal value is denoted as W* and satisfiesSelected by the dynamics equation of the robotUpdate law is
The weight estimation error isDefinition of Filtering errorsCan be obtained from the filter definition
According to different control modes, the moment terms Fa of (24) are respectively:
In the passive control mode
Fa＝-Kpep (29)
Kp is the position gain, ep is the error, and at this time ep＝ep1
In the active control mode, according to the closest point, the unit vectors of the two directions for applying the adjusting moment are found as follows:
Fac＝-(n+b)/||n+b||＝-ep/||ep|| (30)
Ftr=t (31)
n and b represent the normal vector and the secondary normal vector, respectively, at the nearest position, ep is the error, at which point ep＝ep2, t is the tangent vector at the nearest position.
Thereby establishing the force field as
Fa＝k1ω1Fac+k2ω2Ftr (32)
K1 and k2 are control gains for adjusting the magnitude of the output torque, ω1 and ω2 are the weights of the tangential component and the normal component, respectively, r is a variable for adjusting the relative weights of the two components according to the attitude profile error ep;
selecting Lyapunov function as
Its first derivative is
Substituting control law and using property 2 to obtain
Due toAndThen there is
At the same time
In the passive control mode, substitution (29) to (35) is achieved
Wherein the method comprises the steps ofTaking:
Then there is
By adjusting parameters, lambda1 >0 is guaranteed, and the designed controller is stable, so that the system has robustness.
In the active control mode, get
Substituting into (45) to (35) to obtain
Wherein the method comprises the steps ofAnd because of
Substituted into (46) to obtain
Also, since the error ep is bounded, i.e., meets ep||≤ep, Kp needs to meetTaking out
Then there is
The designed controller is stable by adjusting parameters to ensure lambda2 to be more than 0, so that the system has robustness, and the whole block diagram of the control system is shown in figure 4.