Path planning on SE(3) for AUVs using the RRT method
J. Jamieson and J.D. Biggs
https://doi.org/10.19124/ima.2015.001.15
Abstract
This paper presents a kinematic path planning method for a slender autonomous underwater vehicle (AUV) using a technique based on the Rapid-Exploring Random Tree (RRT) algorithm extended to the Euclidean group of motions SE(3). The paths are generated by connecting sub-Riemannian curves on SE(3) that are optimal with respect to a weighted quadratic cost function of the translational and angular velocities subject to the kinematic constraints of the vehicle and are C1 smooth for translational motions. Unlike the majority of other AUV path planning methods, our framework provides a full description of the rotational and translational position as a function of time and the curves satisfy the differential constraints of the vehicle. The modified RRT method is demonstrated in two scenarios, the first with a sparse obstacle field and the second with a dense obstacle field. In both cases the algorithm successfully generates a six degree-of-freedom path from the initial position to a desired final position.