3.1. Machine Certainty-Equivalent Adaptive Control
Generally, the number of constraints on system outputs mainly depends on the control mission and the environment conditions. When
, only
n constraints are activated according to (
2), while when
, augment constraints
with a sufficient large constant
can be added by the designer to the system until
. Without loss of generality, we devise the automatic controller for the scenario with
. The Euler–Lagrange system (
1) with the coordinate
is characterized by (
3), and the corresponding automatic controller
can be reformulated as
where
. A new variable
is defined by
to remove the constraint on
, where
is the reference signal for
with
defined by
and
,
,
. It is noted that
is the projected feasible reference that satisfies all constraints, i.e.,
. Reference signal (
7) is one possible smooth solution. In addition, according to (
7), we also find that
when
m is sufficiently small, indicating that the reference signal tracks the human operator’s intentions as long as the human behavior is sufficiently safe (i.e.,
is far away from obstacles).
Since the smooth function for all , then and exist and , , . Furthermore, note that .
Suppose
denotes a point in
at
space, so the projection of
into the safe subset
with regard to the
ith set of motivation constraints, denoted by
, is characterized by
. Thus, the projection of
into the subset
with regard to the
ith group of active constraints is characterized as
The velocity error is defined by
under the
ith set of motivation constraints. Next, by differentiating
and
, the tracking dynamics of the constrained Euler–Lagrange system is obtained by
where
,
In the backstepping control design framework, the virtual input of the first subsystem in (
9) is formulated as
, where
. Then the virtual input derivative is
. Furthermore, define
, and the tracking dynamics (
9) is reformulated by
where
.
Define
with a tunable constant
; then, the model (
10) can be reformulated as
where
, with the parametric vector
formulated from the terms
,
,
, and
and with the corresponding regression matrix
of
. Actually, this is the result of the linear parameterization property of model (
1). Different sequences of parameters in the parameter vector result in different regression matrices. Thus, the explicit form of the known regressor
vs. the unknown parameter
in (
11) cannot be directly given based on the general model (
1), unless the model (
1) of the mechanical system is explicit.
Design the certainty-equivalent adaptive state feedback controller as
where
K is a positive-definite matrix;
is the estimation of
; and
is a tunable constant.
Theorem 1.
For the nonlinear constrained system (1) under Assumptions 1 and 2, if the state-feedback adaptive control law is designed by (12), then it can be readily demonstrated such that the controlled system satisfies and as . Proof. Choose a candidate Lyapunov function for the system (
11) by
with parametric estimation error
. Then, the derivative of
along the trajectory of (
11) with controller (
12) satisfies
. Since
and
K are negative definite and positive definite matrices, then
, which indicates that both
and
are bounded. In addition, the second derivative with respect to
is given as
. According to the definition of
and (
10), it can be concluded that both
and
exist and are bounded; therefore,
exists and is bounded. From Barbalat’s Lemma [
13], it is immediately concluded that
and
as
. Furthermore, it can be concluded that
and
as
. □
Remark 2.
The concept of certainty equivalence in traditional adaptive control is an important feature. This implies that initially, a controller is constructed under the assumption that all the system parameters are known. Controller parameters are defined by a function of all model parameters. Given the actual values of the model parameters, the controller parameters of the controller are computed by solving design equations aimed at pole-zero placement, model matching, or optimality. When actual model parameters are unavailable, controller parameters can be either estimated directly or computed indirectly by solving the identical design equations utilizing estimations of the model parameters. The obtained controller, which is estimated or designed specifically for the estimated model, is termed a certainty equivalence adaptive controller. However, since certainty-equivalent adaptive law is generally just an integrator about system states, the parameter estimator performance depends on the state tracking performance.
3.3. Machine Adaptive Output-Feedback Control
The error Equation (
9) can be reformulated as
where
,
. Clearly, since
can be written as
based on the linear parametrization, then the model (
23) can be parameterized as
Furthermore, it can be seen that and are the first-order continuously differentiable functions.
The adaptive output feedback control law for (
23) is proposed as
where
and
are filtered signals;
is the estimation of
;
is an auxiliary variable used in the filter;
is a constant, diagonal, positive-definite matrix;
is a positive constant; and
for any
with
being the standard signum function.
To facilitate the stability analysis of the controlled system, define a signal as
Then, taking the time derivative of
in (
27) with a signal in (
30) leads to
Taking the time derivative of
in (
30) yields
After multiplying (
32) by
and substituting (
25) into (
24), the closed-loop system is
where
and the parameter estimation error is
.
Remark 4.
Since defined in (23) is continuously differentiable, it is shown that in (33) is upper-bounded by with , and the positive function is non-decreasing in . Lemma 1.
Define a function . If the control gain matrix is selected to satisfy the sufficient condition , where is the jth entry of the diagonal matrix , then it can be derived that with the positive constant .
Proof. From (
26) and (
30), one has
with
. Then
. □
Theorem 3.
Consider the output-constrained system (1) under Assumptions 1–2. If the adaptive output-feedback controller is designed by (25), then the closed-loop system signals are bounded and and as , where is chosen to satisfy , and with constant . Proof. Define a function
from Lemma 1. Choose a function as
Note that (
34) can be bounded by
where
,
and
. The time derivative of
along the system trajectory of (
33) is given by
where
,
, and
. This means that
for
or
, where
is a positive constant. Then, following Theorem 8.4 in [
13], one can define the region as
. From (
35) and (
37), we know that
,
,
,
, and
are bounded. From (
30), we further know that
is also bounded. Then, from (
26)–(
28), we know that
,
, and
are bounded. Furthermore, since
is a first-order continuously differentiable function, then it is also a bounded function. Then, from (
25) and (
23), we know that the feedback control inputs
and
are bounded. Based on the above boundedness statements, it can be deduced that
is also bounded, which is a sufficient condition for
being uniformly continuous. If we define the region
, then we can recall Theorem 8.4 in [
13] to state that
as
for any
. Then, from the definition of
in Remark 1, we know
,
,
, and
as
for any
. Furthermore, it can be concluded from (
30) that
as
for any
. Therefore, all closed-loop system signals are bounded, and
and
as
for any
. It can be further noticed that larger
results in a larger attraction region
to include any initial conditions of systems so that a semi-global type of stability result is derived. Then, based on
and
, we can derive that the attraction region is
such that
. □
In spite of the availability of the proposed adaptive output-feedback controller (
25), the velocity
of system (
23) is still unknown, so the safety of system (
1) cannot be known in real time. Fortunately, based on the separation principle of observation and control, the velocity observer for (
23) could be designed under the adaptive output-feedback controller (
25). In fact, once the velocity observation
is derived by the velocity observer, the observation for actual velocity
in (
5) can be directly computed from (
6) and the first sub-equation in (
9) as
where
and
. Thus, the objective of velocity observer design is to ensure the observation error
as
only using measurement
.
To facilitate designing the velocity observer for (
23) subject to parametric uncertainty, it is assumed that the matrix
can be written as
with known part
and unknown part
. Then the system model (
23) is rewritten as
where
and
.
Assumption 3.
The parametric uncertainty is bounded, and is a first-order continuously differentiable bounded function in (39). Based on the above assumption, the following velocity observer is designated by
where
is an auxiliary variable and
are constant, diagonal, and positive definite matrices. Furthermore, the observer (
40) can be reformulated as
Then, based on the system model (
39), the observation error system is
Define a signal
and set
with
identity matrix
. Then, from (
42), one has
where
. Note that
and
are bounded functions due to Assumption 3.
It is easily proven that the proposed velocity observer (
40) for system (
23) can guarantee that
and
as
if the gain matrix
satisfies
, where
and
are the
jth entry of
and
, respectively.
Theorem 4.
Given the output-constrained system (1) under Assumptions 1 and 2 and reference signal (7), assume for all . If the adaptive state-feedback controller for (1) is given by (12) or (13) or if the adaptive output-feedback controller for (1) is given by (25), then and as , and for all . Proof. It has been clearly proven in Theorems 1 and 2 that
and
as
under the condition that the feedback controller has no switching from
to
under
. Now consider the case of switching from
to
under
. If
in (
1) switches without delay from
to
at
; then,
and
at
. If
in (
1) switches not directly from
to
, such as if
departs from
at
and falls into
at
, then owing to the existence of
, there is
ensuring that
for
. Define
and note that
; then, we have
and
. Let
represent a succession of motivation constraint groups of which
for
, and the
jth constraint group is motivated in the time
under
with
. Thus,
. Set the multiple Lyapunov function
for whole system if
, then
and
as
for
. Furthermore, founded on the signal
, it can be derived that
; thus,
for
. □
3.4. Human Control Inputs
In order to provide human inputs by joystick handles, the position of handles should be away from its zero position, and the human can give the mechanical systems different speed and direction commands based on the human’s intent. When this occurs, a control force feedback is created according to the human’s hand, where the force from the handles scales linearly with the displacement away from the tracked position along each degree of freedom of the mechanical systems, and the control force can be expressed as
where
is a spring constant and
represents the deviation from the tracked position along each degree of freedom. Note that only the stiffness parameter
is applied to calculate the joystick feedback forces, ensuring that the human undergoes the identical stiffness response in both directions.
Assumption 4.
In general, the human–machine interactive system based on the joystick handles is passive, that is , where denotes the input force exerted by a human, denotes an external force exerted by the environment, is the speed of the handle of a joystick, and is the speed of the controlled mechanical systems. As the mechanical system is not manipulating its environment, then it is observed that no external forces are applied to the system except to the disturbances so that and the human control input is passive.
Measures of the human’s intent to control the position of the mechanical system can be defined by using the above assumption. Denote the rates of the joystick handles displaced from the tracked position as
. Then, the intent of human’s operation is measured as
Note that whenever the joystick handle is moved away from the tracked position in any direction, and only when so that the human’s intent to control the system is measured conveniently.