N00014-15-1-2595
August 1, 2015 – July 31, 2018
$482,492
ONR
Ocean Sensing and Systems Application Division
PI:
S. Ferrari
Project Description
This project develops information-driven navigation and control algorithms for underwater target detection and identification. To this end, information value functions and computational geometry techniques are developed for images beyond the visible spectrum, such as underwater sonar imaging.
Research Goals
- Develop integrated probabilistic model of vehicle motion and side-scan sonar measurements
- Derive information value function for integrated probablistic model
- Develop adaptive motion planning for imaging sonar, so as to optimize a tradeoff of sensing and navigation objectives.
- Demonstrate energy efficient, minimum-time, and other optimal trajectories on one or more sonar-equipped UUVs in variable underwater environments.
Peer-Reviewed Publications
- S.Chang, J. Isaacs, B. Fu, J. Shin, P. Zhu, and S. Ferrari, “Confidence Level Estimation in Multi-target Classification Problems,” SPIE Defense + Commercial Sensing, April 2018. [PDF]
- P. Zhu, J. Isaacs, B. Fu, and S. Ferrari, “Deep Learning Feature Extraction for Target Recognition and Classification in Underwater Sonar Images,” Proc. of the IEEE Conference on Decision and Control (CDC), December 2017. [PDF]
Presentations
- “Deep Learning Feature Extraction for Target Recognition and Classification in Underwater Sonar Images,” IEEE Conference on Decision and Control (CDC), Melbourne, Australia, December 2017 [PDF]
- “Information-driven Guidance and Control for Adaptive Target Detection and Classification” ONR Mine Warfare Autonomy Virtual Program Review, September 7, 2017 [PDF]