Time Preference for Information in Multi-agent Exploration with Limited Communication

Multi-agent exploration of unknown environments with limited communication is a rapidly emerging area of research with applications including surveying and robotic rescue. Quantifying different approaches is tricky, with different schemes favouring one parameter of the exploration, such as the total time of exploring 90% of the environment, at the expense of another parameter, like the rate of information update at a base station. In this paper we present a novel approach to this problem, in which agents choose their actions based on the time preference of the base station for information, which it encodes as the desired minimum ratio of base station utility to total agent utility. We then show that our approach performs competitively with existing exploration algorithms while offering additional flexibility, and holds the promise for much improvement regarding incorporation of various information preferences for the base station.