A Data-Driven Approach for Collision Risk Early Warning in Vessel Encounter Situations Using Attention-BiLSTM

Jie Ma,Chunwei Zhang,Xin Yang,Xiaochun Cheng,Chengfeng Jia,Wenkai Li

doi:10.1109/access.2020.3031722

Abstract

Collision risk early warning is critical to sailing safety in vessel encounter situations because it provides ship officers with sufficient time to react to emergencies and take evasive actions in advance. In this study, we take spatiotemporal motion behaviors of encountering vessels into account since vessel motion behaviors have great influences on the occurrence of a dangerous situation. For this purpose, a data-driven approach is proposed to associate the motion behaviors with the future risk and early prediction of risk is achieved through classifying the behaviors into corresponding risk level. Specifically, we first derive a sequence of relative motion features between encountering vessels to characterize the spatial interactions that vary over time. Then a novel deep learning architecture, which combines bidirectional long short-term memory (BiLSTM) and attention mechanism, is developed to capture the spatial-temporal dependences of behaviors as well as their impacts on future risk. In particular, the BiLSTM is able to discover correlations among behaviors and the attention mechanism can emphasize the key information relevant to the risk prediction task. Exploiting the advantages of these two mechanisms makes the risk prediction more reasonable and reliable. Extensive experiments using ship trace data from the Yangtze River Estuary demonstrate that the proposed Attention-BiLSTM approach outperforms conventional LSTM in terms of accuracy and stability. Moreover, the real-time capability of the approach gives it a significant potential for use in predicting collisions at the early stages.

Highlights

Maritime transportation is the major mode of transportation for world trade
It is useful to distinguish the contributions of various motion behavioral features using weight coefficients. This approach has been shown to improve risk prediction accuracy. To impose such a weighting, we propose a deep learning architecture based on the application of an attention mechanism to the output of the bidirectional long short-term memory (LSTM) (BiLSTM), which is widely used in the field of natural language processing
The bidirectional long short-term memory (BiLSTM) is used to capture the temporal dependence of motion behavior, and the attention mechanism is used to establish a more precise collision risk prediction model based on the varying influence of motion behavior on risk

Summary

Introduction

Maritime transportation is the major mode of transportation for world trade. The continued growth of the global economy has increased the need for ships with larger cargo-carrying capacities and faster sailing speeds. The resulting increases in heavy traffic flow have made maritime accidents a complex problem in most waterways [1]. In busy water areas with dense traffic, ship-ship collisions are the most. Frequently occurring accidents and account for nearly 60% of all maritime incidents [2]. To prevent collisions and improve navigational safety, a variety of risk assessment models, including accident frequency [3], accident consequence estimation [4] and probability estimation models [5]) have been extensively studied. Most of the models developed to date fail to incorporate methods for the early warning of collision risk and instead have tended to focus on the instantaneous assessment of collision risk.

Objectives

Methods

Results

Conclusion