Prev-Talk | Next-Talk | All-Talks | Talks-Sorted

Talk-16: PlumeSim: A Simulation Approach to Autonomous Adaptive Plume Tracking with Multiple AUVs

Stephanie Petillo, MIT, Laboratory for Autonomous Marine Sensing Systems, Center for Ocean Engineering

Building upon the theme of autonomous, adaptive feature tracking with AUVs, last year we presented some field testing of adaptive thermocline tracking with AUVs, and recently we have been working to develop plume detection algorithms for plume tracking. Motivated by the Deepwater Horizon oil spill in 2010 and resulting plumes of oil found suspended at depth in the Gulf of Mexico, as well as frequent harmful algal blooms along our coasts, it has become evident that an efficient and intelligent means of tracking plumes is necessary for monitoring the health of the ocean and coastal environment. At this time, we have developed a simulation environment in which a network of AUVs is tasked with sampling a plume along its boundary at a fixed time slice, and, using that collection of boundary sample points, estimates the plume shape as a sum of Fourier orders. Some challenging topics in this application include spatiotemporal data aliasing (coverage vs. resolution), underwater communication, AUV autonomy, data fusion, and coordination of multiple AUVs. An extension of this work will include time variations of the plume shape and using temporally distributed AUV data to forecast plume development, as well as future implementation with autonomy behaviors controlling the AUVs.

Categories:

  • Plume Tracking
  • Ocean Modeling
  • Ocean/Autonomy Simulation
  • Unmanned Underwater Vehicles (UUVs) / Autonomous Underwater Vehicles (AUVs)
  • Autonomy / Collaborative Autonomy