Roboat

Abstract

With increasing challenges in aquatic navigation, real-time monitoring, and environmental task automation, Autonomous Surface Vehicles (ASVs) emerge as viable platforms for research and real-world applications. This project presents the design, construction, and testing of a low-cost, embedded ASV system capable of performing core navigational and operational tasks aligned with the RoboBoat 2025 competition. The ASV is engineered to autonomously complete six major tasks: buoy channel navigation, obstacle avoidance, speed trials, object detection and retrieval, precise docking, and return-to-home routing. The vehicle integrates GPS-based navigation, a Pixhawk flight controller for PID-based motor control, and a Raspberry Pi 5 for visual input processing using the YOLOv5 object detection framework. A custom-built dual-hull design ensures enhanced stability, while thruster angles and propeller configurations were experimentally tuned for optimal hydrodynamic performance. Challenges such as waterproofing, stability in wave disturbances, and motor calibration were systematically addressed through simulation and iterative testing. This ASV prototype showcases the practical viability of embedded control systems for autonomous marine applications without reliance on high-end AI or ROS frameworks.