Replication Data for: Risk-based path planning for preventing collisions and groundings of maritime autonomous surface ships

Field trials on risk-based path planning with an autonomous surface vehicle.
Two field trial experiments were performed to test and demonstrate risk-based path planning with KPRA, a method for dynamic probabilistic risk assessment, and risk evaluation on an autonomous surface vehicle (ASV) from Maritime Robotics. The main objective was to observe the effect of KPRA on an ASV’s behavior, compared to its nominal behavior, in an uncontrolled and uncertain environment. The first experiment tested KPRA’s effect on the ASV navigating open waters during a crossing encounter with another vessel. The second experiment increased the environmental complexity and tested the method in a crossing encounter near an island. Both experiments yielded datasets, called \"open_waters\" and \"near_land\" respectively, containing data on the ASV's state (e.g., position, speed and heading), the other vessel's state, and the explored planning state-space (i.e., the states considered by KPRA when searching for an adequate path), among others. Both experiments were performed in areas around Munkholmen island in Trondheim, Norway. The island is a tourist spot in Trondheim, frequently approached by recreational vessels and kayaks, so the risks of collision and grounding are present. In both experiments, the main goal for the ASV was to autonomously navigate near the island while performing risk assessment and avoiding undesired consequences - in this case, collision with other ships and powered groundings.