06.02.2013 Views

Underwater Robots - Gianluca Antonelli.pdf

Underwater Robots - Gianluca Antonelli.pdf

Underwater Robots - Gianluca Antonelli.pdf

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

7.8 Output Feedback Control 169<br />

C T ( q , σ d ) ζ o + D ( q , ζ ) ζ − 1<br />

2 D ( q , ζ r )(ζ r + ζ o ) . (7.50)<br />

Astate vector for the closed-loop system (7.49),(7.50) isthen<br />

⎡<br />

⎢<br />

x = ⎣<br />

σ d<br />

e d<br />

σ e<br />

e e<br />

⎤<br />

⎥<br />

⎦ .<br />

Notice that perfect tracking of the desired motion together with exact estimate<br />

of the system velocities results in x = 0 .Therefore, the controlobjective<br />

is fulfilled if the closed loop system (7.49), (7.50) is asymptotically stable at<br />

the origin of its state space. This is ensured by the following theorem:<br />

Theorem. There exists achoice ofthe controller gains K p , K v , Λ d and of<br />

the observer parameters L p , L v , Λ e such that the origin of the state space of<br />

system (7.49),(7.50) islocally exponentially stable.<br />

Consider the positive definite Lyapunov function candidate<br />

V = 1<br />

2 σ T d M ( q ) σ d + 1<br />

+ 1<br />

2 k pP �η<br />

T<br />

B<br />

1 ,d<br />

2 σ T e M ( q ) σ e +<br />

�<br />

B<br />

�η 1 ,d + k pO (1 − �η d ) 2 + �ε T<br />

d �ε d<br />

�<br />

B<br />

�η 1 ,e + l pO (1 − �η e ) 2 + �ε T<br />

e �ε �<br />

e<br />

�<br />

+ 1 T<br />

�q d K pQ�q d +<br />

2<br />

+ 1<br />

2 l T<br />

B<br />

pP �η 1 ,e + 1 T<br />

�q e L pQ�q e . (7.51)<br />

2<br />

The time derivative of V along the trajectories ofthe closed-loop system<br />

(7.49),(7.50) isgiven by<br />

˙V = − σ T d K v σ d − e T deΛ d K p e d − e T e Λ e L p e e +<br />

− σ T e<br />

�<br />

L v A ( � �<br />

Q e ) − K v σ e − σ T d C ( q , σ e ) ζ r +<br />

− σ T e C ( q , ζ ) σ e + σ T e C T ( q , σ d ) ζ o +<br />

+(σ d + σ e ) T D ( q , ζ ) ζ + − 1<br />

2 ( σ d + σ e ) T D ( q , ζ r )(ζ r + ζ o )+<br />

− σ T d M ( q ) S PO( �ν 2 ,d) ζ d − σ T d M ( q ) Λ d S P ( �ν 2 ,d) e de . (7.52)<br />

In the following it is assumed that �η d > 0, �η e > 0; in view of theangle/axis<br />

interpretation of the unit quaternion, the above assumption corresponds to<br />

considering orientation errors characterized by angular displacements inthe<br />

range ]− π, π [.<br />

From the equality � Q de = � − 1 Q e ∗ � Q d ,the following equality follows<br />

�ε eT<br />

de �ε d = �η e �ε T<br />

d �ε d − �η d �ε T<br />

d �ε e ,<br />

where �η d and �η e are the scalar parts of the quaternions � Q d and � Q e ,respectively.<br />

The above equation, in view of �η B<br />

1 ,de = �η B<br />

1 ,d − �η B<br />

1 ,e and �q de = �q d − �q e ,<br />

implies that

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!