Vision of the Seas: Open Visual Perception Framework for Autonomous Sailing Vessels

We propose an Open Visual Perception framework for autonomous navigation and obstacle detection in marine sailing vessels. The proposed framework lays the groundwork for perceiving marine objects needed for vessel navigation and obstacle avoidance using only image processing as the primary method. Various methods are explored to extract features such as waterline, terrain, and other vessels. The proposed framework also incorporates image stabilization techniques without the need for an Inertial Measurement Unit to compensate for ocean wave oscillations. The work also introduces the system architecture of an autonomous marine vessel, besides the components of the open visual perception framework. Results demonstrate the capabilities of obstacle detection, horizon line identification, and water texture detection.