Algorithm for Determining the Position of a Ship Hull-Cleaning Robot

Abstract

The paper explores the development and evaluation of algorithms for the positioning of ship hull cleaning robots, focusing on machine learning and sensor fusion techniques. The research employs Gradient Boosting, Kalman filters, and deep learning to enhance the accuracy of robot positioning. Gradient Boosting is used to predict displacement vectors and rotation angles, while the Kalman filter is applied to refine position estimates by integrating odometry and GPS data. Deep learning models are utilized to predict robot trajectories based on sensor inputs. Experiments conducted on the Rosario dataset and simulated environments demonstrate the effectiveness of these methods.