Tracking and Predicting Barge Locations on Inland Waterways

Abstract

In efforts to improve efficiency, safety, and security, several groups involved in inland waterway navigation have an interest in improving tracking of barges moving on waterways. Automated tracking devices have inherent limitations, and there is a need to predict locations over the next several hours. This paper presents a nonlinear, probabilistic prediction model developed and implemented to track spatial location and other navigation characteristics of a barge traveling on the inland waterway system. A prefilter to check the validity of the measurements, a nonlinear speed estimation process, and a Kalman filter to predict the navigation solution of the barge are developed in this work. Due to the complex dynamics involved in the barge navigation system, a nonlinear stochastic model was developed in state space to represent the process and measurement processes. The algorithm was verified using actual measurements obtained from multiple barges on multiple rivers acquired from different sensors. The results show a reliable and robust prediction algorithm for tracking inland waterway barges.