A system for the validation of collision avoidance algorithm performance of autonomous ships

Adequate validations on intelligent ships are essential to guarantee navigation safety and efficiency. In this paper an emulated validation system is constructed to test collision avoidance ability of autonomous ships in open waters, coastal regions and archipelago areas with different traffic densities. A synthetic map generator is designed, the map meshing conbines the techniques from Voronoi Tessellation, Poisson-disk Sampling and Perlin Noise to build up more natural coastlines and detailed elevation data. To imitate random encounter scenarios, a new method is designed to maintain target ship quantity, to conduct various tests, 3 types of water geography, different ship behaviors as well as manual and automatic control modes are applied, during testing, each ship is designed to follow the planned routes which are generated arbitrarily and then smoothed. A validation scheme is put forward for comprehensive testing procedures. Finally, A set of experiments on the collision avoidance al-gorithm show the merits and drawbacks, this verifies the effectiveness of the validation system.