The purpose of this project is to research and develop the first stage of an Obstacle Avoidance System (OAS) for the Robotic Oceanographic Surface Sampler Unmanned Surface Vehicle (ROSS USV). At this stage, the OAS can detect obstacles in front of the vessel and report back to the operator their position relative to the USV. The system operates on a low-powered Intel NUC computer on board the ROSS vessel. Images are collected from a camera mounted on the front of the ROSS and are stabilized using data from an Inertial Measurement unit (IMU) sensor. A SSD neural network pre-trained on the Singapore Maritime Dataset is used to identify objects visible in captured images. Once detected, objects are added to a tracking system. The tracking system allows for operators to uniquely identify obstacles, and roughly measures how the distance between the ROSS and tracked obstacles changes over time. The OAS reports detected objects to the host computer which notifies the operator of possible hazards in front of the ROSS in real time.