Knot The Average Boat: Development of a Semi-Autonomous Sailing Platform

Abstract

The problem of autonomous sailing is a complex and challenging one. There are nearly countless possible environmental and telemetry sensors that can be implemented, each with varying challenges and benefits, and similarly many possible implementations of feedback control and navigation. This project approached the problem from a bottom-up approach, first creating a sensory and computational architecture that met capability requirements, then subsequently building a programmatic feedback control system designed to maximally leverage the capabilities of the onboard electronics hardware. The sailing platform is a 50-inch model sloop collecting wind data from a waterproof rotary encoder equipped with a wind vane. The onboard microcontroller, a Raspberry Pi Pico, has three autonomy modes: Manual, Pilot-Assist, and Autonomous Tack, which runs a closed-loop feedback control system with full control over rudder and sail servos. The feedback controller was designed with real-world dynamics data and leverages discrete-time controller simulations to verify reliability. The highest level of autonomy reached in this project is considered to be semi-autonomous; whereas all servo commands were autonomously controlled, the platform nonetheless requires an external operator to send a command for when to execute turns. Once completed, the vessel was fully tested in real-world conditions in the presence of highly variable winds and performed admirably, demonstrating a robust and stable controller for semi-autonomous tacking.