06.02.2013 Views

Underwater Robots - Gianluca Antonelli.pdf

Underwater Robots - Gianluca Antonelli.pdf

Underwater Robots - Gianluca Antonelli.pdf

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

88 4. Fault Detection/Tolerance Strategies for AUVs and ROVs<br />

4.5 Experiments<br />

Roby 2isaROV developed at the National Research Council-ISSIA, Italy.<br />

It has been object of several wet tests aiming at validating fault detection<br />

approaches. The horizontal motion isobtained bythe use of two fore thrusters<br />

that control the surge velocity aswell as the vehicle heading; the depth<br />

is regulated by means of two vertical thrusters. In[3, 4, 5] experiments of<br />

different fault detection schemes have been carried out by causing, on purpose,<br />

an actuator failure: one ofthe thrusters has been simply turned off. The<br />

experiments in [3, 4] have been carried out in apool, in [5] the experiments<br />

also concern acomparison between EKFs and sliding-mode observers. The<br />

latter is aresult of abilateral project with the Naval Postgraduate School,<br />

Monterey, CA.<br />

The Italian National Research Council (CNR-ISSIA) also developed the<br />

ROV ROMEO and tested, in an antarctic mission, both fault detection and<br />

tolerant schemes [52, 61]. In particular the case of flooded and blocked thrusters<br />

occurred. Inboth cases the fault has been detected and the information<br />

could be reported tothe human operator in order to activate the reconfiguration<br />

procedure. Figure 4.7 shows the expected and measured motor currents<br />

in case of flooded thruster: itcan be observed apersistent mismatching between<br />

the output ofthe model and the measured values.<br />

The vehicle Theseus manufactured by ISE Research Ltd with the Canadian<br />

Department ofNational Defence successfully handled afailure during<br />

an Arctic mission of cable laying [117, 118]. In details, the vehicle did not<br />

terminate ahoming step, probably due to poor acoustic conditions and the<br />

Fault Manager activated asafe behavior: stop under the ice and wait for further<br />

instructions. This allowed tore-establish acoustic telemetry and surface<br />

tracking and safely recover the vehicle. Notice that his fault wasn’t intentionally<br />

caused [118].<br />

The vehicle ODIN, anAUV developed at the Autonomous Systems Laboratory<br />

(ASL) of the University ofHawaii, HI, USA, has been used for several<br />

experiments. In [306, 307] the fault detection and tolerant schemes are experimentally<br />

validated. The thruster fault hasbeen tested by zeroing the output<br />

voltagebymeans of software, the fault detection schemeidentified the trouble<br />

and correctly reconfigured the force allocation byproperly modifying of the<br />

TCM. The fault tolerant scheme with respect to depth sensor fault has also<br />

been tested by zeroing the sensor reading and verifying that the algorithm,<br />

after aprogrammed time of 1s,correctly switched on the other sensor. While<br />

the theory has been developed for a6-DOFs vehicle, the experiments results<br />

only present the vehicle depth.<br />

The same vehicle has been used to validate the fault tolerant approach<br />

developed in [232, 233, 251] in 6-DOFs experiments. Different experiments<br />

have been carried out by zeroing the voltage onone or two thrusters simultaneously<br />

that, however, did not cause the vehicle becoming under-actuated.<br />

The implemented control law isbased onanidentified reduced ODIN model

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

Saved successfully!

Ooh no, something went wrong!