Incorporation of delayed decision making into stochastic mapping

Abstract: This paper presents a technique for incorporating delayed decision mak-ing into stochastic mapping algorithms for concurrent mapping and localization. The approach explicitly tracks the error correlations between current and previous vehi-cle states, enabling the initialization of map features using data from multiple time steps and improved data association decision-making. The method is illustrated us-ing data from a ring of Polaroid sonar sensors from a B21 mobile robot, demonstrat-ing the ability to perform CML with sparse and ambiguous data. 1