Grid Based Sensing Model for Autonomous Ship Agent

The cost of computation becomes a bottleneck for an agent based marine traffic simulation. This is especially true when investigating interferences of a large number of ships and static objects placed in the sea. In general, a collection of ship agents built with a simple sensing model mandates a quadratic time complexity of O (n^2) for sensing each other. The runtime cost required for sensing the environment sharply increases and the system performance rapidly deteriorates, as the number of environment elements increases. This paper presents a grid based sensing model, a new approach for building an autonomous ship agent. The aim of this research is to address the issue of time and space complexities. The proposed model reduces the cost of time and space for running the marine traffic simulation system implemented with the previous work of a ship cluster behavior model. Analysis of the grid based sensing model and its overall effects based on simulation experiments are discussed.