Proving theSafety ofMotion Estimation Algorithms: aConfidence Interval Approach
Massar Marc, Olivier Bilenne, Druet Emmanuel · 2006
Thispaperwillpresent innovative algorithms designed forprovingtheintegrity ofparameter estimation. These algorithms rely onanunusual sensor modeling, delivering measurements as confidence intervals (with anassociated confidence level) rather thanaspoint estimates, asusually considered. Suitable combinations ofmeasurements provided byseveral sensors mayreach extremely highconfidence levels for theestimate, andmaythusserve estimation problems wheresafety isofuttermost importance. Onetypical application ismotion estimation forautomated guided transport (railways forinstance), butourconfidenceinterval basedalgorithms aregeneral enough toapply forother safety-demanding applications.