Time-bounded mission planning in time-varying domains with semi-MDPS and Gaussian processes

Uncertain, time-varying dynamic environments are ubiquitous in real world robotics. We propose an online planning framework to address time-bounded missions under time-varying dynamics, where those dynamics affect the duration and outcome of actions. We pose such problems as semi-Markov decision processes, where actions have a duration distributed according to an a priori unknown time-varying function. Our approach maintains a belief over this function, and time is propagated through a discrete search tree that efficiently maintains a subset of reachable states. We show improved mission performance on a marine vehicle simulator acting under real-world spatio-temporal ocean currents, and demonstrate the ability to solve co-safe linear temporal logic problems, which are more complex than the reachability problems tackled in previous approaches