energy-based control of under-actuated mechanical systems - remotely driven acrobot
TRANSCRIPT
![Page 1: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/1.jpg)
Energy-Based Control of Under-Actuated
Mechanical Systems: Remotely Driven Acrobot
By Xin Xin, and Taiga Yamasaki
Presentation by Mostafa Shokrian Zeini
![Page 2: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/2.jpg)
Important Questions:
- What is an under-actuated mechanical system (UMS)? And what are its applications?
- What is the energy-based control?
- How to provide a global motion analysis of the UMS under the designed controller?
- What are the conditions on control parameters for achieving a successful swing-up control?
2
![Page 3: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/3.jpg)
Under-Actuated Mechanical Systems
Under-actuated Mechanical Systems (UMSs) possess fewer actuators than degrees of freedom.
The design of such mechanisms allows to reduce cost and weight by using fewer actuators and permits to increase fault tolerance to actuator failure.
Controlling these systems is challenging due to inherently complex nonlinear dynamics.
So
However
3
![Page 4: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/4.jpg)
Under-Actuated Mechanical Systems
Cart-Pendulu
m System
4
![Page 5: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/5.jpg)
Under-Actuated Mechanical Systems
Translational
Oscillators with
Rotating Actuator (TORA)
5
![Page 6: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/6.jpg)
Under-Actuated Mechanical Systems
Gyroscope
6
![Page 7: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/7.jpg)
Under-Actuated Mechanical Systems
Pendubot&
Acrobot
This is a two-link planar robot with a single actuator and with a first joint being attached to a passive joint.
7
![Page 8: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/8.jpg)
Acrobots
Acrobots
The Directly Driven Acrobot (DDA)
The second link is directly driven.
The Remotely Driven Acrobot
(RDA)
The second link is remotely driven by
an actuator mounted at a fixed base through a belt.
8
![Page 9: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/9.jpg)
Acrobots
9
![Page 10: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/10.jpg)
Acrobot
The motion equations
10
![Page 11: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/11.jpg)
Directly Driven Acrobot (DDA)
The motion equations
11
![Page 12: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/12.jpg)
The motion equations
Directly Driven Acrobot (DDA)
12
![Page 13: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/13.jpg)
Remotely Driven Acrobot (RDA)
The motion equations
For our RDA, the angle of link 2 with respect to the –axis, i.e. , is actuated via the timing belt:
13
![Page 14: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/14.jpg)
Remotely Driven Acrobot (RDA)
The motion equations
14
![Page 15: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/15.jpg)
Remotely Driven Acrobot (RDA)
The motion
equations
15
![Page 16: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/16.jpg)
Remotely Driven Acrobot (RDA)
The motion
equations
Regarding the RDA motion equations with respect to , and the DDA described with respect to , although their control input transformation matrices are the same, the other coefficient matrices are different. Thus, by no means the RDA motion equations are like the DDA dynamics.
16
![Page 17: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/17.jpg)
Energy-Based Swing-Up Controller
where is the potential energy of the RDA expressed as:
𝑃 (𝜃 )=𝛽1 (cos𝜃1 −1 )+𝛽2 (cos𝜃2−1 )
The total mechanical energy of the RDA is given by:
The control objective is to find a controller under which the RDA can be swung up:from any initial state to any small neighborhood of the Upright Equilibrium Point
(UEP)
17
![Page 18: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/18.jpg)
where the constant is a given reference of and satisfies
𝐸𝑟 ≥ min𝐸=−2(𝛽1+𝛽2)
In what follows, our goal is to design a controller such that:
lim𝑡→ ∞
𝐸=𝐸𝑟 , lim𝑡 → ∞
𝜃2=0 , lim𝑡→ ∞
�̇�2=0
Energy-Based Swing-Up Controller
18
![Page 19: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/19.jpg)
Lemma 1
Suppose constant control parameters , , and satisfy:
, and , respectively, then
contains no singular points. Moreover, as the closed-loop solution converges to the largest invariant set at which and where and are constants; and the following equations hold
19
Energy-Based Swing-Up Controller
![Page 20: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/20.jpg)
Energy-Based Swing-Up Controller
If can be chosen such that where :
Taking the time derivative of , together with the property yields
Consider the following Lyapunov function candidate:
20
𝑴 𝜽 (𝜽 ) �̈�+𝑯 𝜽 (𝜽 , �̇� )+𝑮𝜽 (𝜽 , �̇� )=𝑩𝑨𝝉
![Page 21: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/21.jpg)
Energy-Based Swing-Up Controller
Under the condition and using
Putting it into gives:
We obtain :
21
𝑴 𝜽 (𝜽 ) �̈�+𝑯 𝜽 (𝜽 , �̇� )+𝑮𝜽 (𝜽 , �̇� )=𝑩𝑨𝝉
![Page 22: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/22.jpg)
Energy-Based Swing-Up Controller
yields every approaches as .
Thus, and are constant in .
Let be the largest invariant set in . Using:
Yields:
22
𝑴 𝜽 (𝜽 ) �̈�+𝑯 𝜽 (𝜽 , �̇� )+𝑮𝜽 (𝜽 , �̇� )=𝑩𝑨𝝉
It proves that the controller contains no singular points.
![Page 23: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/23.jpg)
Energy-Based Swing-Up Controller
If , then is a constant denoted as . Putting and into the above equation yields:
Putting and into :
From , is also constant in : This proves
23
𝑴 𝜽 (𝜽 ) �̈�+𝑯 𝜽 (𝜽 , �̇� )+𝑮𝜽 (𝜽 , �̇� )=𝑩𝑨𝝉
This completes the proof of lemma 1.
![Page 24: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/24.jpg)
Energy-Based Swing-Up Controller
Therefore, since the (C2) holds for all , the following relation must hold:
where is constant. Since , is bounded: So the energy and the angular velocity are bounded.
Integrating it with respect to time yields:
24
𝑴 𝜽 (𝜽 ) �̈�+𝑯 𝜽 (𝜽 , �̇� )+𝑮𝜽 (𝜽 , �̇� )=𝑩𝑨𝝉
![Page 25: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/25.jpg)
Energy-Based Swing-Up Controller
where is constant. Similarly, . Thus:
is a constant noted as .
Integrating the above equation with respect to time yields:
Consequently:
25
𝑴 𝜽 (𝜽 ) �̈�+𝑯 𝜽 (𝜽 , �̇� )+𝑮𝜽 (𝜽 , �̇� )=𝑩𝑨𝝉
![Page 26: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/26.jpg)
Energy-Based Swing-Up Controller
Now, Let’s bring the Lemma 2.
Putting and into (C1) gives:
Putting and into the above equation yields , thus:
26
𝑴 𝜽 (𝜽 ) �̈�+𝑯 𝜽 (𝜽 , �̇� )+𝑮𝜽 (𝜽 , �̇� )=𝑩𝑨𝝉
![Page 27: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/27.jpg)
Lemma 2
Suppose constant control parameters , , and satisfy:
, and , respectively, and
If the convergent value of energy satisfies , then the closed-loop solution converges to an equilibrium point satisfying
27
Global Motion Analysis of RDA
![Page 28: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/28.jpg)
Let’s consider two situations:i- ii-
From (C1) : .
Thus by using :
28
i-
Global Motion Analysis of RDA
![Page 29: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/29.jpg)
We recall the result of Lemma 2:
We want to find conditions such that could be the unique solution of the above equation.
Clearly, is a solution of the above equation.But if , …
29
ii-
Global Motion Analysis of RDA
ii-
![Page 30: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/30.jpg)
Thus, is the unique solution of , if and only if:
Putting into yields , where:
30
Global Motion Analysis of RDA
![Page 31: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/31.jpg)
The closed-loop solution of system converges either to the UEP or the down-up EP.
Let’s Introduce the Theorem 1.
Under the above condition, is the unique solution of And according to the result of Lemma 2 ()
where
31
Global Motion Analysis of RDA
![Page 32: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/32.jpg)
Theorem 1Consider the closed-loop system with positive parameters , , and . Suppose that If , and control parameter satisfies
then under the designed controller, for any initial condition of RDA, either of the following two statements hold:i) , and the closed loop solution converges to ii) , and the closed loop solution converges either to the UEP or the down-up EP .
32
Global Motion Analysis of RDA
![Page 33: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/33.jpg)
Theorem 2
Consider the closed-loop system with positive parameters , , and . Suppose that . Suppose that . The following statements hold:i) The Jacobian matrix of evaluated at down-up EP has 2 and 2 eigenvalues in the open left- and right-half planes, respectively; such a down-up EP is unstable.ii) The Jacobian matrix of evaluated at UEP has 3 and 1 eigenvalues in the open left- and right-half planes, respectively; such an UEP is an unstable.
33
Global Motion Analysis of RDA
![Page 34: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/34.jpg)
The closed-loop solution converges to:
The closed-loop solution converges to the down-up EP which is unstable.
34
ii)
Global Motion Analysis of RDA
With the quantities described in Theorem 1, if :
i)
![Page 35: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/35.jpg)
Cosidering , starting from almost all initial state:
the closed-loop solution approaches as .
Since the down-up EP is unstable:the RDA under the designed controller cannot maintain at the this point
in practice.
35
Global Motion Analysis of RDA
![Page 36: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/36.jpg)
36
Global Motion Analysis of RDA
![Page 37: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/37.jpg)
The linearized model of the RDA around the UEP is:
where and:
37
Global Motion Analysis of RDA
is controllable and one can obtain a locally stabilizing controller: , where is the state-feedback gain.
![Page 38: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/38.jpg)
Defining the following small neighborhood of the UEP:
where is a given positive weighting matrix, and is a given small positive number.
When the RDA is swung up into the neighborhood of the UEP, we can switch the swing-up controller to the locally stabilizing controller.
38
Global Motion Analysis of RDA
![Page 39: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/39.jpg)
39
Simulation Results
The switch was taken about at which the RDA was driven into the neighborhood of the UEP.
![Page 40: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/40.jpg)
40
Simulation Results
![Page 41: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/41.jpg)
41
Simulation Results
![Page 42: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/42.jpg)
42
Simulation Results
![Page 43: Energy-Based Control of Under-Actuated Mechanical Systems - Remotely Driven Acrobot](https://reader037.vdocuments.net/reader037/viewer/2022102903/55caa444bb61eb27688b46df/html5/thumbnails/43.jpg)
References
1. X. Xin, T. Yamasaki, “Energy-Based Swing-Up Control for a Remotely Driven Acrobot - Theoretical and Experimental Resultsˮ, 2012, IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, 20 (4), pp. 1048-1056.
2. X. Xin, S. Tanaka, J. She, T. Yamasaki, “New Analytical Results of Energy-Based Swing-up Control for the Pendubotˮ, 2013, International Journal of Non-Linear Mechanics, 52, pp. 110-118.
3. N. Adhikary, C. Mahanta, “Integral Backstepping Sliding Mode Control for Underactuated Systems - Swing-up and Stabilization of the Cart–Pendulum Systemˮ, 2013, ISA Transactions, 52, pp. 870-880.
4. B. Gao, H. Song, J. Zhao, C. Gong, “Dynamics and Energy-based Control of TORA System on a Slopeˮ, 2013, Proceeding of the IEEE Int. Conf. on Cyber Tech. in Automation, Control and Intelligent Systems, Nanjing, China, pp. 373-378.
5. M. Pourmahmoud Aghababa, H. Pourmahmoud Aghababa, “Chaos Synchronization of Gyroscopes using an Adaptive Robust Finite-time Controllerˮ, 2013, Journal of Mechanical Science and Technology, 27 (3), pp. 909-916. 43