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

Talk-01: Receding Horizon Planning and Tracking for Monitoring Surface Vessels with Passive Sonar: Initial Field Experiments

Speaker: James Turner, James McMahon, Philip D. Baldoni, Benjamin R. Dzikowicz, Naval Research Laboratory (NRL) USA

Autonomous underwater vehicles (AUVs) must maintain situational awareness of nearby surface vessels in order to safely come to the surface. For some applications, active sonar may be unsafe to use, requiring the vehicle to rely solely on passive sonar to sense nearby vessels. However, passive sonar provides much less information, so accurately estimating the states of the vessels is more challenging.

With passive sonar, the measurement at a given time consists of a set of relative bearings to possible acoustic sources, i.e. targets. The ranges of the targets must be inferred from the sequence of bearing-only measurement sets received over time. Movement of the AUV, which is carrying the sensor, is critical: if the sensor is stationary, a close, slow-moving target and a distant, fast-moving target produce equivalent measurement sequences. Some movements of the AUV are more useful than others, since they place it in a position and orientation to receive measurements which provide more new information. So, planning the trajectory of the AUV given the current knowledge of the targets can improve the future estimates of the targets' states.

In this talk, we describe our approach for (1) estimating the states of nearby surface vessels from bearing-only measurements and (2) planning the motion of the AUV, according to the current belief, to maximize the information gain from future measurements. Additionally, we discuss how we've integrated this combined filter and planner into MOOS-IvP, the issues we experienced during our initial in-water testing, and how we plan to resolve them.

Categories:

  • AUVs/UUVs
  • Sensing/Perception