Underwater Robots - Gianluca Antonelli.pdf
Underwater Robots - Gianluca Antonelli.pdf
Underwater Robots - Gianluca Antonelli.pdf
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
52 3. Dynamic Control of 6-DOF AUVs<br />
• ˙ V ( s v , ˜ θ v ) ≤ 0<br />
• ˙ V ( s v , ˜ θ v )isuniformly continuous<br />
then<br />
• ˙ V ( s v , ˜ θ v ) → 0as t →∞.<br />
Thus s v → 0 as t →∞.Inview of the definition of s v ,this implies that<br />
˜ν → 0 as t →∞;inaddition, due to the properties of the quaternion, it<br />
results that ˜η → 1as t →∞.However, as usual inadaptive control schemes,<br />
it is not possible to prove asymptotic stability ofthe whole state since ˜ θ v is<br />
only guaranteed to be bounded.<br />
Notice that, in [121], further discussion is developed by considering different<br />
choices for the matrix Λ .<br />
In 1997 [89] G.Conte and A.Serrani develop aLyapunov-based control<br />
for AUVs. The designed controller, by introducing arepresentation ofthe<br />
model uncertainties, ismade robust using Lyapunov techniques. It is worth<br />
noticing that this approach does not take into account explicitly for an adaptive/integral<br />
action tocompensate for the current effect. The position error is<br />
represented within avehicle-fixed representation with afeedback term similar<br />
to the vector s v used inthis Section.<br />
Compensation ofthe persistent effects. Following the same reasoning<br />
as previously done for the law A ,itcan be deducted that also in this case<br />
the adaptive compensation cannot guarantee null position error at rest in<br />
the presence of ocean current. In fact, the dynamic model (2.51) shows that<br />
asteady null linear velocity ofthe vehicle with anon-null position error<br />
does not excite acorrective adaptive control action. Again, ocean current<br />
measurement cannot overcome this problem inpractice.<br />
To achieve anull position error in presence of ocean current, an integral<br />
action on body-fixed-frame error variables was considered in[34, 35]. However,<br />
this integral action has the following drawback: let suppose that the<br />
vehicle is at rest inthe left configuration of Figure 3.2 in presence ofawater<br />
current aligned to the earth-fixed y axis; the control action builds the current<br />
compensation term which, in this particular configuration, turns out to be<br />
parallel to the y b axis. If the vehicle is now quickly rotated toright configuration,<br />
the built compensation term rotates together with the vehicle keeping<br />
its alignment tothe y b axis; however, this vehicle-fixed axis has now become<br />
parallel to the x axis ofthe earth-fixed frame. Therefore, the built compensation<br />
term acts as adisturbance until the integral action has re-built proper<br />
current compensation for the right configuration. It is clear at this point that<br />
this drawback does not arise for the controller E and for the controllers A<br />
and B ,since they build ocean current compensation in an earth-fixed frame.<br />
Reduced Controller. Similarly to the other cases, areduced form of the<br />
controller has been derived. An integral action on body-fixed-frame error