Initial Results of Data Acquisition for Constructing a Multimodal Maritime Object Dataset

Abstract

This paper presents a multimodal sensor dataset for maritime object detection and tracking in commercial shipping routes, addressing the limited availability of such data due to legal restrictions. Collected at Ulsan Port, an international hub, the dataset provides simultaneous lidar, radar, camera, GNSS, and IMU time-series data for maritime objects at distances ranging from 100 meters to 10 kilometers. The dataset allows researchers and developers to explore object detection algorithms for camera images, radar images, and lidar point clouds, while also testing sensor fusion and tracking algorithms and evaluating their performance against Automatic Identification System (AIS) data. While this dataset contributes to the availability of comprehensive sensor data, future work includes calibration of cameras, lidar, and radar, data collection for complete arrival and departure scenarios, and deployment in formats other than rosbags. By providing this resource, we hope to support innovation and facilitate the development of maritime systems that can contribute to the safety and efficiency of commercial shipping operations.