Autonomous Navigation on US (Electronic) Nautical Charts

TitleAutonomous Navigation on US (Electronic) Nautical Charts
Publication TypeConference Proceedings
Year2018
AuthorsSchmidt, VE
EditorReed, S
Conference Name2018 Canadian Hydrographic Conference
Conference DatesMarc 26-29
PublisherCanadian Hydrographic Society
Conference LocationVictoria, CA
KeywordsASVs, ENCs, nautical charts, unmanned systems

Although much interest has been given to the use of autonomous surface vehicles (ASVs) for hydrographic data collection, little thought has been given to the utility of currently available chart products for safe navigation of the ASV itself. In the United States, chart products are currently available in digital form, as both cartographic raster images of traditional paper charts and as vector representations of cartographic data, (“BSB” files and electronic nautical charts (ENCs), respectively). Here we evaluate these chart products with an eye to common methods by which artificial intelligence (AI) algorithms would likely use them. We find that the raster cartographic nature of BSB nautical charts leaves a complex interpretation problem for computers to recognize and understand their nuances. However, the BSB cartographic representation holds useful information that can be equally difficult to infer from electrical nautical charts, particularly when the size of objects are implicitly tied to the scale of the chart. Further, we find that while ENCs provide near instantaneous interpretation, the data must be reorganized for fast search. Additionally, some features, notably docks and breakwaters, are represented in the ENC in a single dimension (a line) even though they subtend a finite second dimension, forcing the AI algorithm to buffer objects to ensure safe navigation. When objects fail to have explicit measurements (for example a measured depth) encoded in the ENC, one is left to interpret their relative hazard from qualitative descriptions. This interpretation can be particularly challenging when the qualitative descriptions are referenced to the local vertical datum. Finally, the ENC’s compilation scale, when encoded, is particularly useful as it provides an implicit measure of uncertainty about the chart information, determining the granularity with which navigation choices can be made.