Simulation Framework and Software Environment for Evaluating Automatic Ship Collision Avoidance Algorithms^*

Research on maritime automatic collision avoidance algorithms is nowadays abundant, and becoming more mature. Moreover, methods for the evaluation of such algorithms are emerging. Therefore, we propose a software environment for simulation and an evaluation framework for maritime collision avoidance, used for testing and evaluating control algorithms. The system comes with a set of standard scenarios that can be augmented with random variations and a database of scenarios based on historical Automatic Identification System (AIS) data that can be used in the simulation. In this framework, the user can set up scenarios with pre-programmed target ship routes, or with target ships that exhibit proactive and reactive collision avoidance maneuvers. Moreover, the user can specify uncertainty in the tracking of target ships and situational awareness, and include the presence of grounding hazards from Electronic Navigational Charts (ENCs). The evaluation tool computes penalties and scores on algorithm performance according to the maritime traffic rules (COLREGS), safety, and other performance-related metrics. The framework can be used for verification and assurance, and further enables automatic adjustment of collision avoidance planners online or offline.