A comprehensive spatiotemporal metric for ship collision risk assessment

Collision risk assessment is an effective way to determine the potential collision and reserve time for the action taken to avoid the possible collision. This paper investigates the ship collision risk assessment problem from a new perspective that considers both the spatial and temporal factors related to the ship's navigational safety. A spatiotemporal ship domain (SD) model is developed to take both factors into account so that any traditional 2-dimensional SD model can be generalized into a 3-dimensional spatiotemporal SD model. Based on the new SD model, a comprehensive spatiotemporal metric is introduced to predict the ongoing ship collision risk, which considers not only the collision probability but also the assumed consequence. Moreover, the new assessment algorithm is continuous, and the quantitative collision risk can be obtained. Compared to the previous SD methods, the advantage of the new algorithm is that it can provide a more comprehensive and flexible risk assessment when facing different encounter situations. The effectiveness of the new risk assessment algorithm is examined by simulation studies which assesses the collision risk for a real ship collision accident and show advantages in comparison with the previous methods. •The ship collision risk assessment problem was investigated from a new perspective that considered both the spatial and temporal factors related to the ship’s navigational safety.•A spatiotemporal ship domain model was developed to integrate the spatial and temporal attributes into one comprehensive model.•A geometric probability model which used the spatiotemporal distance and the volume of the overlapped region between the spatiotemporal ship domains was proposed to evaluate the collision risk.•The algorithm also provided a continuous and quantitative collision risk assessment with respect to the states of ships.•Three encounter situations, i.e. head-on, crossing and overtaking encounter situations, were considered in the simulations to examine the effectiveness of the new assessment algorithm.