Fast Trajectory Generation on a Path using Feedback Linearization*

Abstract

This paper describes a method for quickly generating feasible trajectories for fully actuated autonomous surface vehicles. It is assumed that a path consisting of a list of positions with associated heading angles is given. The goal is to generate a feasible trajectory that follows a given path and takes into account the nonlinear dynamical model. The trajectory contains the state as well as force-level input variables and can be used as feed-forward control by a trajectory controller. The generation of the trajectory is based on a closed-loop simulation of the nonlinear model in combination with a nonlinear controller and guidance law. Feedback linearization is applied to linearize the motion model of the surface vehicle at the velocity level. A state feedback controller is used to control the pose along the path. The presented method allows for simple and fast trajectory generation and can therefore be applied in sampling-based motion planners where many trajectories have to be generated.