In this paper, we address the problem of partial reconstruction of underwater landscape in a region using a group of formations of autonomous cooperating vehicles. Sampling is adaptive in that only a small set of iso-contours are sampled, as opposed to the more traditional strategy of sampling along pre-defined trajectories sweeping unnecessary areas. To achieve this, we model the environment as a tesselation of the plane by triangular elements with linear interpolation inside each element. Each formation consists of three cooperating vehicles which traverse the tesselation element-wise by computing ascent (descent) directions using the simplex algorithm and climbing to reach a specific isocline. We extend the standard simplex algorithm to the case of moving over isocontours.

Paper (pdf) | other publications