Sensing for Hydrographic Autonomous Surface Vehicles

TitleSensing for Hydrographic Autonomous Surface Vehicles
Publication TypeConference Proceedings
Year2019
AuthorsMoreno, C, Schmidt, VE, Calder, BR, Mayer, LA
Conference NameU.S. Hydrographic Conference (US HYDRO)
Conference DatesMarch 19-21
PublisherThe Hydrographic Society of America
Conference LocationBiloxi, MS
Keywordsasv, Autonomous Survey Systems, Sensor Fusion

With increasing interest in the use of autonomous surface vessels (ASVs) to automate hydrographic data collection in support of safe navigation, there is a growing likelihood that ASVs will be operated in regions with uncertain or limited prior knowledge of where it is safe to navigate. In addition to this challenge, coastal environments may have significant boat traffic and other hazards for an unmanned vehicle, such as buoys, lobster pots and kelp. If ASVs are to operate safely and truly autonomously, means must be developed to increase the awareness of the ASV to its environment so that it can safely maneuver with minimal operator intervention.

This paper provides a review of a variety of sensing systems used by ASVs for the identification of obstacles on the surface and underwater, their detection and classification capabilities, and their limitations and uncertainties. AIS, radar, LiDAR, color and infrared (FLIR) cameras, multibeam echo-sounders and forward looking sonar are considered. The paper will explore how to use the complementary nature of these sensors in order to offer the best possible environmental perception and situational awareness. In addition, the paper will look at a number of obstacle types, evaluate their detection requirements, and match these requirements with the sensors available aboard the ASV, including the determination of which sensors provide actionable information natively, and which require further algorithm development. Finally, the use of deep learning algorithms will be examined using data from the ASV’s camera to better understand the applicability of deep learning for the detection of objects in the marine environment.

Check When Not CCOM Publication