Context
This project modeled a single-agent autonomous underwater vehicle exploring an unknown constrained cave environment. The system used Stonefish and ROS2 to simulate underwater dynamics, sensor feedback, and asynchronous autonomy modules, with PLUME-generated cave geometry providing the exploration environment. The autonomy stack integrated sonar-based perception, SLAM, frontier detection, 3D A* path planning, waypoint tracking, and formal runtime monitoring to evaluate both mission progress and safety constraints.
What I contributed
I helped design and implement the autonomy stack, focusing on the planning, control, and safety-verification layers. My work included modeling the 6-DOF AUV under sensing and actuation constraints, programming frontier-based waypoint generation over a partially observed occupancy grid, and integrating path-planning behavior with waypoint tracking.