Trajectory planning for motion-constrained AUVs in uncertain environments

Providing a series of images of a site over time with a survey-class AUV presents numerous challenges, particularly in the process of getting close to rugged terrain with a motion-constrained vehicle in an uncertain environment. To deal with this, a baseline approach presented in previous work by the authors plans spline-based trajectories based on an a priori map of the terrain, allowing for improved performance over purely reactive control schemes. This paper extends that approach to account for uncertainty in the environment, both in the knowledge of the terrain and the motion of the vehicle, providing additional robustness and safety, while minimizing the potential loss in performance. The trajectory optimization approach is demonstrated over simulated terrain.