Past Event
We are a vendor-neutral, non-profit 501(c)(3) organization comprised of members from industry, government, and academia worldwide. Our mission is to educate and connect the HPC & AI user community with the latest technologies and best practices to optimize business processes and promote workforce advancement.
We focus on the high-performance systems used for AI, simulation, cloud computing, quantum computing, and visualization. Our members utilize these technologies in their work in oil & gas, renewable energy, climate modeling, life sciences, manufacturing, financial services, cybersecurity, government, and academia.
The Society of HPC Professionals past
Lunch & Learn Round Table Discussion Meeting
Lunch & Learn – March 2019
Deep Learning Applications in Manned Spaceflight
Held 28 March 2019
Download presentation:
[swpm_protected for=”2″ format_protected_msg=”1″]
[/swpm_protected]
Abstract:
Human spaceflight is deeply rooted in close collaboration between Mission Control and flight crew. Ground control acts like a black box converting vehicle telemetry, crew natural language input, and 2D video into time management and problem-solving decisions, significantly contributing to crew and vehicle efficacy by delegating cognitive load.
As next-generation missions take humans deeper into the solar system, signal propagation delays threaten to hamper this effectively real-time cognitive tether. Deep learning technologies running locally aboard spacecraft can produce intelligent systems capable of making inference-based decisions like Mission Control without the time delay, while also augmenting crew-computer interfaces in ways not seen before.
This talk will discuss three applications leveraging deep learning on a first step toward this goal. The first is an Intelligent Personal Coach app for exercise feedback. The second is a neural network to identify external vehicle handrails in 2D images to assist crew members in identifying touch points suffering Micro Meteroid and Orbital Debris (MMOD) damage which threaten spacesuit structural integrity.
Finally, a method to generate rendered image data sets for training a neural network to identify 6 degree-of-freedom pose (used by an augmented reality procedure assistant for alignment without a QR code) will be presented.
Speaker Bio:
Matthew Noyes
Software Lead at NASA’s Hybrid Reality Advanced Operational Concepts Lab
Matthew Noyes is the lead software engineer and deputy lab manager of the NASA Johnson Space Center Advanced Operational Concepts Lab in Houston, Texas. His topics of research include building and deploying advanced technologies leveraging machine learning and augmented reality in future space missions to augment crew safety, human performance, and vehicle systems management.
Location:
The University of Houston Classroom and Business Building (CBB)
Room 522, 4742 Calhoun Rd.