Equipment Support for Cloud-based Intelligent VR Systems
Abstract
We request support for purchasing essential instrumentation to pursue novel research related to modeling and simulation of complex virtual environments using machine learning algorithms in the Cloud. The underlying research will lay the scientific foundation of a new area related to ÒIntelligent VR SystemsÓ, where machine-learning and AI algorithms are used to automatically reason, extract, and analyze the real physical world for modeling and construction of virtual environments with intelligent systems. Furthermore, the high-end workstations with multiple CPUs and GPUs will also support our current research on computational acoustics, physics-based simulation, and intelligent agents (including both virtual humans and social robots). This equipment will also support the ongoing research under the following funded projects by the U.S. Army Research Office (ARO) and Office of the Undersecretary of Defense for Intelligence and Security (OUSDI), and will be used in our joint collaborative projects with Army Research Laboratory (ARL), United State Army Corps of Engineers (USACE), and Institute of Creative Technologies (ICT). 1. Efficient Computational Models for Simulating Large-Scale Heterogeneous Crowds (PI: M. Lin) 2. Acoustic Simulation in Real-World Scenes (PI: D. Manocha) 3. Understanding the Commercial Landscape for Insider Threat Detection (PI: D. Manocha) 4. Real-time Architecture for Emotion and Behavior Insights from Real-time Gait-Analysis (PI: D. Manocha) 5. Social and Behavioral Science Research on Insider Threat (PI: D. Manocha) In terms of equipment support, we propose to acquire high-performance parallel computing capabilities for cloud computing, in support of creation of intelligent VR systems. More specifically, we request two different types of mixed computing platforms, as they have different capabilities in terms of networking and storage capacities Ð high-end multicore Lambda Workstations and Lambda GPU servers for cloud computing for studying and analyzing workload issues between CPU-GPU clusters. Similarly, we request to acquire three types of multi-agent systems, including eight Turtlebots, four Rover Robots, and three Social Robots. Each one of them offers different kind of sensing and locomotion capabilities that capture various forms of intelligence embodies by these social robots. They will be used to train and test intelligent behaviors using deep learning models for perception, navigation and human-system interaction. Besides multi-human and intelligent system interactions in the physical and virtual worlds, the techniques developed using the proposed equipment will also be useful for ongoing research projects on computational acoustics, physics-based simulation, virtual reality and artificial intelligence.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 04, 2021
- Source ID
- W911NF2110026
Entities
People
- Ming C. Lin
Organizations
- Army Contracting Command
- United States Army
- University of Maryland