Design of an Automated Obstacle Detection System for Collision Early Warning on a Small Fishing Vessel

Indonesia has a high potency for fisheries commodities, produced by not only large and medium sized fishing vessels, but also traditional and small fishing vessels. Most traditional and small fishing vessels are not equipped with adequate navigation tools, although the wider range navigation devices are not a mandatory for small fising vessel, remaining the increasing an accident risk. The National Transportation Safety Committee of Indonesia (KNKT) reported approximately 31% of ship accidents occur on fishing vessel from 2018 to 2020, with around 342 people missing and dying in such accidents, including ship collision accidents. To address that issue, this study developes a deep learning-based obstacle detection system to improve the safety of traditional and small fishing vessels, which are often involved in maritime accidents. In this study, YOLOv8 algorithm is used to detect four types of obstacle objects: fishing vessels, tugboats, dredging vessels, and cargo ships which are assumed to be a common involved collided ship againts the small fishing vessels. The obstacle detection system is then embedded into a single board computer unit as central processing unit for collision detection. The best object detection training model was produced after 120 iterations, showing a high level of accuracy with a precision of 87.1% and a recall of 81.6%. This system is firstly tested by using a recorded video representing a video-captured an onboard camera, continuing with a direct testing on a small fishing vessel. Both implementations confirm that this system can be used for detecting an obstacle in front of the vessels as well as provides early warning to prevent a collision. By applying this system as a collision early warning system on a small fising vessel, this system is expected to overcome real-world needs in reducing the risk of fishing vessel casualties, enhancing maritime safety, and supporting the digital transformation of the national fisheries sector.