iGSDI Awards AI+ Rapid Innovation for Defense and Security Grants
May 15, 2018
AI+ Rapid Innovation for Defense and Security Winners
Autonomous Exploration and Localization of Targets for Aerial Drones
Dr. David Cappelleri, Associate Professor, Mechanical Engineering
Dr. Eugenio Culurciello, Associate Professor, Electrical and Biomedical Engineering
There is a growing need in the DoD and Army to fight unconventional weapons and resilient fighters in war zones. These can be in the form of snipers hidden in buildings and also small remote-controlled drones that carry weapons and explosives.
We propose to develop algorithms based on deep learning and computer vision to identify targets, and map areas, in particular indoor and in foliage and large vegetation. The algorithms will be able to fly a small drone to find specific target of interests in the area, while providing a comprehensive exploration of the area or building.
Intelligent Biomorphic Robots with Adversarial Machine Learning Capacity
Dr. Xinyan Deng, Associate Professor, Mechanical Engineering
Dr. Bowei Xi, Associate Professor, Statistics
The team will build intelligent combat biomorphic robots with adversarial machine learning capacity, with the ability to detect mobile robots from moving animals in natural settings, while protecting itself from been detected by adapting to learned biological locomotion principles as unique camouflage. Such robots are uniquely suited for dealing with challenging real life situations in the battlefield.
An AI-based Hybrid Pilot Drowsiness Detection System
Dr. Dengfeng Sun, Associate Professor, Aeronautics and Astronautics
Dr. Xiao Wang, Professor, Statistics
In this proposal, we develop an AI-based hybrid drowsiness detection system, which is a self-powered system providing visual or aural warnings and real-time feedback to a pilot. This research aims to integrate monitoring and estimation of pilot fatigue and stress level through the development and application of deep learning algorithms, combined with a novel data fusion technique that incorporates both aircraft dynamics and situational awareness.
Automatic Target Recognition for Weapon Seeker Target Acquisition or Re-acquisition from Unmanned Aerial Vehicles (UAVs)
Dr. Juan P Wachs, Associate Professor, Industrial Engineering
Dr. Dong Hye Ye, Research Assistant Professor, Electrical and Computer Engineering
Dr. Charles Bouman, Professor, Electrical and Computer Engineering
The objective of this task is to develop an automatic target recognition system to allow a seeker to acquire or reacquire the target form the video taken by unmanned aerial vehicles (UAVs). We assume a moving target and the goal is real-time shape and trajectory learning. For real-time learning, it is not necessary to re-train the classifier for every new target; especially since only a limited dataset is available. This approach allows for a compact representation of the target and allows the use of a deep learning based edge detector to bridge the gap between the predicted signature and actual data.