Autonomous decision-making is key to safe and efficient marine autonomy, as global marine industry comprises over 90 percent of the world's cargo transportation. Challenges of the real-world validation in the aquatic domain limits the wide-spread of ASVs despite their promising societal impacts. We propose and demonstrate the real-world validation platform and comprehensive algorithm steps. Such a framework will serve as a more explainable and reliable decision-making system of ASVs as well as autonomous vehicles in other domains.
We would like to thank Prof. Eun-Bang Lee and Youngsoo Park in Korea Maritime and Ocean University for help with application of marine traffic data and collision avoidance. This work is supported in part by the Burke Research Initiation Award and NSF CNS-1919647, 2144624, OIA1923004.