High-Throughput Computing and Multi-Sensor Unmanned Aerial Systems in Support of Explosive Hazard Detection
Abstract
This proposal is a two pronged request for high throughput computing (HTC) and multi-sensor unmanned aerial system (UAS) equipment in support of active U.S. Army Research Office (ARO) grants on machine learning and fusion for explosive hazard detection (EHD) (W911NF-18-1-0153, W911NF-17-1-0193, W911NF-10-1-0279) and an Army Research Laboratory (ARL) cooperative R&D agreement (CRADA) on wide area motion imagery (WAMI). Investigators at the University of Missouri-Columbia have produced algorithms for the U.S. Army on EHD and WAMI within and across sensors and platforms for over 30 years. Whereas we have excellent resources, we lack the infrastructure capabilities outlined herein to make a leap (theory and application) and conduct more detailed, realistic, and repeatable platform/sensor/target experiments on past, current, and future Army data. Our HTC request is for data storage, a high-end server for compute-intensive offline experimentation, a development server, and mobile edge-computing for real-time UAS activities. The mobile edge-computing equipment is designed to also be used in an offline capacity for additional day-to-day compute resources. Our UAS/sensor request focuses on new U.S. Army capabilities. In EHD, existing platforms (e.g., hand-held, forward-looking, and side-looking (we have active research in all)) unfortunately put a human in harmÕs way; in close or direct proximity to threats. A UAS can mimic existing platforms and keep the human at a safe stand-off distance. Our UAS request consists of two copters and a fixed wing plane. These platforms allow us to investigate potential and best practices in collecting and processing data relative to metrics like task completion time/area covered. Our sensor request includes infrared, high resolution color, lidar, multispectral cameras, and positional sensors. These sensors have/are used in our Army research for WAMI and/or non-aerial EHD (hand-held and ground vehicle). Some sensors provide a direct way to sense and detect objects, e.g., explosive hazards. The remaining sensors facilitate registration of multi-sensor data and/or they can be used for scene understanding/context/tracking/etc. In summary, the U.S. Army will see advancements in EHD and WAMI as a result of this DURIP. In addition, the various PIs/Co-PIs and Army projects involve overlapping topics like signal/image processing, registration, fusion, machine learning, etc. As such, the Army will get a value-add since this DURIP will bring together investigators/projects and allow for cross pollination of ideas and codes. Our team has demonstrated its ability to perform state-of-the-art research for the Army.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 22, 2019
- Source ID
- W911NF1910181
Entities
People
- Derek T. Anderson
Organizations
- Army Contracting Command
- United States Army
- University of Missouri