Prev-Talk | Next-Talk | All-Talks | Talks-Sorted | MOOS-DAWG'22

Talk-22: Adaptive Depth Contour Following Behavior

Christopher Gagner NUWC Division Newport

Scott Sideleau, NUWC Division Newport

Michael J. Walsh, NUWC Division Newport

Michael L. Incze, NUWC Division Newport (retired)

Maritime robotics platforms equipped with Mission Oriented Operating Suite (MOOS) and Interval Programming (IvP) Helm have been employed successfully by academia and public sector research institutions to achieve various applied research goals over the last 20-years. Often, reactive behaviors that respond to through-the-sensor data have been exercised and refined. This work highlights a specific behavior - adaptive depth contour following - as it has been developed and deployed by NUWC Newport to support various operational demonstrations with the US Navy over the last 12-years. In this update to our on-going case study, we explore the evolution of the behavior from a brute force approach anchored in single ping bathymetric data to its current iteration, which features near-real time bathymetric plane generation and applied regression analysis techniques to improve execution efficiency (i.e. maximize time on-contour). Results obtained in simulation environments and at international experimentation venues using Acoustic Doppler Current Profilers (ADCP) integrated onto Unmanned Undersea Vehicles (UUV) will be discussed alongside lessons learned from evolving the artificial intelligence (AI) approaches used to address this problem.

  • Sensing
  • UUVs