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

Talk-07: Environmentally-Sensitive Search Behaviors for Collaborating Underwater Autonomous Vehicles

Ryan Goldhahn and Kevin LePage, Centre for Maritime Research and Experimentation (CMRE).

The Cooperative ASW Programme at the NATO Centre for Maritime Research and Experimentation is developing multistatic littoral surveillance network concepts that depend on distributed sensing and decision making by autonomous underwater vehicles (AUVs). Since the location of the AUVs dramatically affects the performance of the multi-static autonomous sensor network, optimizing the sensor positions for detection, localization, classification, and/or tracking performance is of primary importance.

This work is focused on adding acoustic propagation models to MOOS-IvP processes and behaviors to facilitate autonomous navigation which takes into account quantities such as range-dependent environmental parameters, anisotropic reverberation, signal/array parameters, and hypothesized target depths. In general, these behaviors calculate the performance of a AUV given by some objective function over a large region, then navigate the AUV along the path which gives the greatest increase in performance, taking into account the location and expected performance of other sensors in the network. More specifically, the anticipated signal excess is calculated using the ARTEMIS propagation model over a grid of hypothesized target and receiver locations. These can be used in path planning to navigate the vehicle along the path which maximizes the cumulative probability of detection over the region. Alternatively, the vehicle can combine these data in a Bayesian framework with in-situ detections in the observed acoustic data and calculate a posterior distribution on target position.

Several mission- and information-based performance metrics are presented to optimize the vehicle path. In the current work, vehicle depth is considered only when a target track is present. The optimal vehicle depth is found which minimizes target transmission loss using the estimated AUV and target positions and the coherent propagation model BELLHOP. When multiple vehicles/receivers are present, navigation solutions are desired which are optimal over the entire network. Collaborative efforts to maximize the chosen objective function over the network are hindered by the limited communication bandwidth between sensors/vehicles. Global optimization methods thus are sought which require minimal sharing of information between platforms. A suboptimal collaborative solution is presented requiring only the position and heading of the collaborating vehicles. These processes are implemented and tested using the MOOS-IvP middleware using the pOctave process also written at the CMRE. Results are shown for both simulation and real data collected during the Co-operative LittoraL Asw Behavior '13 (COLLAB13) sea trial conducted by the CMRE.

Categories:

  • Autonomous Underwater Vehicles (AUVs)
  • Collaborative Autonomy
  • MOOS-IvP
  • IvP Behaviors
  • Anti-Submarine Warfare