Abstract
Infinitesimal look-ahead stopping rules are discussed when the detection capability of a search sensor as well as the existence of an object in a given area for search is not known with certainty. We attempt to utilize "dummies" to obtain extra information which is expected to give us better estimate of the detection capability of the sensor. Stopping rules are investigated for two different criteria; that is, maximizing the expected net return and minimizing the expected search time subject to the condition that the preassigned detection probability of the object is ensured. We first show a sufficient condition under which an infinitesimal look-ahead stopping rule is optimal in the case of the expected net-value criterion. Second, we show that under the above condition, an optimal stopping rule for the expected search-time criterion has the same structure as the expected net-value case. Finally, we discuss the efficiency of utilizing "dummies" by numerical examples.
