Kim, Jonghwi (KAIST), Sol, Han (KAIST), Kim, Jinwhan (KAIST)

Data Association Leveraging Extrinsic Parameter Consistency for Multi-Object Detection in Maritime Environment

Scheduled for presentation during the Regular Session "Underwater Robots" (FrAT4), Friday, July 18, 2025, 10:00−10:20, Room 108

Joint 10th IFAC Symposium on Mechatronic Systems and 14th Symposium on Robotics, July 15-18, 2025, Paris, France

This information is tentative and subject to change. Compiled on August 2, 2025

Abstract

This paper presents a novel data association method utilizing extrinsic parameter consistency, designed to improve the detection and tracking of multiple objects by fusing camera and marineradar measurements. The method addresses the challenges in dynamic maritime environments by leveraging the consistency of extrinsic parameters. To mitigate ambiguities in determining these parameters, a new cost function considering the costs associated with all possible pairs in a given combination is introduced. The constraints related to the alignment and sequencing of measured bearings are also integrated, and dynamic programming is employed to enhance computational efficiency. The enhancements in accuracy and computational speed by the proposed method have been validated using datasets obtained from real-sea field experiments.