The Application of Computational Geometry for Automated UUV Mission Planning

Abstract

UUV survey missions are complicated by environmental factors such as complex bathymetry, seamounts, depth constraints, shorelines, and even entanglement hazards such as wrecks, seaweed, fishing lines, and corals. Missions are also complicated by vehicle performance characteristics and limitations. Often, mission failures are due to planning faults, omissions, and oversights created by human operators while attempting to simultaneously mitigate these factors while contending With less-than-ideal planning conditions such as heat, sand, rain, choppy seas, etc. In this paper, we investigate automating path planning via the application of computational geometry algorithms to reduce the cognitive load on operators during mission planning activities, and accelerate the survey mission planning process. The desired goal is a reduction in mission and vehicle risk, faster overall mission planning which leads to more time actively on mission, and the creation of an automated mission planning pipeline for uncrewed systems. We have focused on the basic problem of automatically generating a set of laivn-moWer patterns that minimize total vehicle travel distance while covering a desired area and avoiding strict exclusion areas (e.g., islands, shoals, seamounts, intakes, or military operations areas). This can be rephrased as the basic computational geometry problem of finding a minimum set of rectangles that span the desired area and do not overlap with exclusion areas. Additionally, we can impose further restrictions such as minimizing coverage area outside the desired area (as in open Water around an island), or requiring no rectangle extend past the boundary of the desired search area (such as in a lake with boundary shorelines). We will describe the process by using well-known computational geometry techniques to automatically generate the set of rectangular regions and hoiv this will lead to the desired outcomes from this work. Further investigation is warranted for arbitrary polygons With more than one arbitrary exclusion region.