Machine Learning-Based Position Prediction for Fleet Management Within Port Terminals

Efficient management of truck fleets within port terminals is crucial for optimizing operations and minimizing turnaround times. This paper explores the potential of machine learning techniques for predicting truck positions. We propose a Long Short-Term Memory (LSTM) network-based approach that utilizes historical time series data of truck movements to predict the next position of trucks within a port terminal. The model addresses the challenge of multipath and Non-Line-of-Sight (NLOS) issues commonly faced with GPS positioning in such environments. We detail the data acquisition process, model architecture with LSTM layers, training procedure, and evaluation metrics used in our research. The trained model achieved a Mean Absolute Error (MAE) of 2.2 m on the validation dataset, demonstrating its effectiveness in predicting truck positions within the acceptable error range specified by port authorities. This research highlights the potential of machine learning for improving truck position prediction accuracy, leading to better planning, resource allocation, and overall operational efficiency within port terminals.