The Autonomous Helmsman: Algorithms and Sensors for Unmanned Littoral Navigation

The Navy is currently developing an Unmanned Surface Vehicle (USV) program. Currently operating systems require human operators for each USV. Manual operation of USVs results in an inefficient use of labor and complex staffing requirements over a long-term deployment. SARA proposes the “Autonomous Helmsman,” a modular system that allows USV navigation according to COLREG Rules and without human intervention. The Helmsman is based on “subsumption architecture”, a behavior-based model derived by robotics expert Rodney Brooks. Subsumption architecture provides a biomimetic decision basis for a state machine, resulting in “reflex” actions (reactive) such as obstacle avoidance and “cognitive” (deliberative) actions such as navigation behaviors. The system allows for evolution of abilities through by an innate ability to subsume new behaviors. The Helmsman realizes behaviors by modifying an initial path, accounting for moving vessels, obstacles, and COLREG Rules. Path modifications are made using an evolution-based path planning algorithm. As part of Phase I, SARA proposes to demonstrate Helmsman operations on our set of unmanned kayaks. Supporting the Helmsman is a localization system and a sensor array capable of detecting obstacles and dynamic vessel characteristics. The proposed trade study identifies the ideal suite of sensors for detecting the environment and vessel attitude.