Hardware Podcasts

21. AI Rescues Firefighters, NASA Prints Spacecraft Parts, Plug & Play Prosthetics

The Next Byte Podcast · 8 Jun 2021 · 23m

AI in practice: one real use of AI described

What this episode covers

Hosts Daniel and Forbode discuss three engineering stories: an AI model by NIST that predicts flashover in fires using sensor data and simulations; NASA’s use of ultrasonic additive manufacturing (UAM) via subcontractor Fibrosnic to create complex, dissimilar-metal heat exchangers; and Levitate, a company founded by an amputee in partnership with the Technical University of Denmark, which offers affordable, plug-and-play athletic prosthetics using proprietary composite materials.

AI in practice

Where this conversation describes AI being used to build or deliver a product. Part of the AI in practice tracker.

Engineering & simulationNIST machine learning model86% accuracy in predicting flashover within one minute

A machine learning model was trained on 5,000 computer simulations of house fires to predict the timing and location of flashovers using data from heat and smoke detectors.

“they actually created a 3D simulation on a computer of a house, and they ran over 5000 simulations on it in different configurations … by tracking the way that the temperature climbed in certain areas of the house, using these temperature sensors, they're able to predict weather flashover happens by doing all these simulations and training a machine learning model.”

Speaker 1 · Hear it at 3:37

Most relevant to

early-stage companies and large enterprises

Topics

Mentioned

Companies
NIST, NASA, Fibrosnic, Levitate, Technical University of Denmark
Tools
Ultrasonic Additive Manufacturing

Related episodes from other shows

Other shows on the same ground, closest first.

  • Additive Snack · 30m

    A.I. in Additive Manufacturing with Omar Fergani of 1000 Kelvin

    Omar Fergani, co-founder and CEO of 1000 Kelvin, explains how their AI-driven platform AMAIZE acts as a co-pilot for additive manufacturing engineers, reducing physical iterations by predicting thermal issues and optimizing print recipes. The company, based in Berlin with US operations, focuses on moving AM from prototyping to serial production by treating it as a materials processing discipline, enabling companies to print existing cast designs without modification. Fergani highlights use cases in oil and gas spare parts, partnerships with EOS and Autodesk, and the necessity of AI to address workforce shortages and scale production.

    AI & machine learningDesign for manufacturing & NPI
    10 chapters

  • Hackaday Podcast · 1h 07m

    Ep 207: Modular Furniture, Plastic Prosthetics, and Your Data on YouTube

    Editors Elliot Williams and Tom Nardi discuss a week of hacks, including a pico balloon shot down by the US Air Force, a NASA SLS rocket noise study, and modular furniture systems like Open Structures. They explore data storage on YouTube and VHS, AI-generated Pong on a Raspberry Pi Pico, and a fake CO2 monitor that is actually an alcohol sensor. The episode highlights Precious Plastic's work in Indonesia creating prosthetic limbs from recycled plastic, Linux audio mixing with PipeWire, and the retro Nintendo Robo.

    Prototyping & iterationMaterials & manufacturing processes
    9 chapters

  • Additive Snack · 54m

    NASA Update: Master Class, Alloys, AI & the Future Engineering Generations

    Paul Gradl, Principal Engineer at NASA, discusses the expansion of additive manufacturing materials, highlighting high-performance alloys like GR COP and GRX 810 developed for extreme environments and now used in semiconductors and energy. He explains how AI and modeling are accelerating alloy development, while NASA shares public data to help the industry overcome qualification challenges. Gradl also details the scaling of large-scale parts via Directed Energy Deposition (DED), the role of additive in space infrastructure including lunar regolith printing, and a recent test stand failure that underscored the importance of understanding process sensitivities like build interruptions.

    Materials & manufacturing processesScaling production
    9 chapters

  • Additive Snack · 44m

    How NASA is Leveraging AM to Accelerate Space Exploration

    Paul Gradl, a principal engineer at NASA, discusses the integration of additive manufacturing (AM) into space exploration, highlighting the transition from early plastic prototypes to flight-qualified metal components like combustion chambers and injectors. He details NASA's collaboration with private companies through data sharing and commercial space act agreements, emphasizing the importance of understanding the entire process chain, including post-processing and certification. Gradl also addresses the lessons learned from field failures, such as witness line issues in GRCop-42 parts, and outlines the future of AM in space, including larger scale printing and in-situ resource utilization on the Moon and Mars.

    Mechanical engineeringManufacturing partners & offshoring
    8 chapters

  • Additive Snack · 1h 11m

    From Hypersonics to AI Workflows: How Ursa Major Is Scaling Rocket Production

    Thomas Pomorski of Ursa Major discusses the company's progress in hypersonics, solid rocket motors, and in-space propulsion, highlighting the successful flight of the 3D-printed Hadley engine and the Draper engine with AFRL. He details Ursa's 'unit cell' manufacturing approach for solid rocket motors to avoid fixed tooling costs and demonstrates how his small team uses AI tools like Claude Code to rapidly prototype slicer features and build data-monitoring web applications. Pomorski explains that while AI significantly boosts development productivity by acting as a force multiplier for experienced engineers, production validation remains a challenge due to the difficulty of verifying AI-generated code and surrogate models.

    AI & machine learningScaling production
    8 chapters

The summary, chapters and topics are generated from the episode’s audio by this directory, not written by The Next Byte Podcast. The episode itself belongs to its publisher.