Policies for risk-aware sensor data collection by mobile agents (Q2151909)

In a certain area \(n\) objects are located which have to be checked by an observer, e.g. a drone. The observer, having a base \(b\), collects the information about the objects on a tour from the base \(b\) through the area with the locations of the objects back to the base \(b\). The locations of the objects and the available connections are modeled by a graph \(G=(V,E)\) with a set \(V\) of nodes, representing the locations of the objects including the base \(b\), and the set \(E\) of out edges describing the connections among locations of the objects. For each information about an object which the observer returns to the base, a reward is received. Moreover, due to risks of destruction, the observer has a probability \(0 <\psi <1\) of survival per unit length of travel. For closed tours \(W\) through the graph, the expected reward is determined resulting from the received rewards and taking into account the probability of possible failures. Characteristics of an optimal tour are considered and optimal tours are derived for special cases. Generalizations to multi-observer problems are discussed, and an example is given.