My talk will explore a general method for modelling trajectories as stochastic differential equations being driven by a potential function. Approximations reduce the task of estimating the potential function to regression-like problems allowing a flexible potential surface to be estimated using readily available tools. The potential function provides a useful summary of movement but also allows the investigation of covariates that may influence movement. Two examples will be used as illustrations: tracks of whale sharks tagged off the Kenyan Coast with archival tags, and the motion of the eye during fixation on a target.