Generation of State Transition Models Using Simulations for Unmanned Sea Surface Vehicle Trajectory Planning

Trajectory planning for unmanned sea surface vehicles (USSVs) in high sea-states is a challenging problem. Large and somewhat stochastic ocean forces can cause significant deviations in the motion of the USSV. Controllers are employed to reject disturbances and get back on the desired trajectory. However, the position uncertainty can be still high and needs to be accounted for during the trajectory planning to circumvent collisions with the obstacles. We model the motion of the USSV as Markov decision process and use a trajectory planning approach based on stochastic dynamic programming. A key component of our approach is the estimation of transition probabilities from one state to another when executing an action. In this paper, we present algorithms to generate state transition model using Monte Carlo simulation of USSV motion. Our simulations are based on potential flow based 6-DOF dynamics. Using this approach, we are able to generate dynamically feasible trajectories for USSVs that exhibit safe behaviors in high sea-states in the vicinity of static obstacles.