COVIA: Computer Vision based Intelligent Assistant for Mistake Proofing of Complex Maintenance Tasks on Navy Ships
Abstract
The purpose of this Naval Engineering Education Consortium (NEEC) proposed project is to investigate advanced deep learning-based computer vision methods and algorithms to enable next-generation Handheld Augmented Reality (HAR) based complex maintenance tasks. Specifically, this proposal asks the following pertinent questions when studying HAR based maintenance systems: How to reliably identify/predict failure modes with limited or reduced data? How one can computationally and efficiently reconfigure HAR workflow process to enable scaling of HAR systems in maintenance workflow? This proposal lays out a rigorous, multi-step computational strategy to answer these questions, which could dramatically change the way we design the next generation HAR based maintenance systems. In order to successfully implement HAR based maintenance systems, two critical challenges need to be addressed: (1) the ease of reconfigurability of the overall system to scale it to different types of product maintenance and (2) leveraging HAR to improve the training of deep learning networks on their ability to diagnose problems. The proposed project also provides an excellent opportunity to engage graduate and undergraduate students in a Navy related research problem and expose them to open-ended, hands-on projects that are more effective in promoting experiential learning. The research focus is well aligned with PH-01 NEEC topic area and will encourage students to consider exciting career opportunities associated with the US Naval Enterprise.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 15, 2021
- Source ID
- N001742110010
Entities
People
- Rahul Rai
Organizations
- Clemson University
- United States Navy