Barron Stone

This has been a really fun project to work on. With thousands of successful simulated drops and hundreds of actual drops, the hard work Drone Amplified put into the new MONTIS is starting to pay off. The new electronic ignition system allows us to detonate a charge at a user-defined height above the snow. The new system is also safer, easier to use, and offers a better cost-per-shot. This video illustrates the versatility of the system, dropping 5- or 10-pound cast boosters and ANFO bags. The MONTIS offers avalanche professionals another tool in their toolbox and will augment the existing, proven tools in use. Proud to work with the team here at Drone Amplified and excited to make snow slide for years to come.

22

Lots of talk about #autonomy and #fpv -- end users asking "Can your drone do XYZ?" Sure it can -- there are a lot of talented devs working on autonomy projects like advanced combined #operations, anti-#radar/counter-#jammer seeking, human-machine #teaming/#swarming. No single company is going to have all the abilities, without working with these #experts in their respective areas of focus. It just needs to be implemented on the right, low cost #hardware so it can meet the need that and be #future proofed for the next #evolution. But what if the hardware was your already cost-effective flight controller, with a few more bells and whistles -- like #open #Linux on-board #compute -- and it simply supported a #Docker (or similar) containerized application that enabled these functions (and new ones folks haven't thought of yet)? What if your FPV wasn't just an FPV but a real "#multirole #multicopter" (that's what Orqa FPV "#MRM" means). Check out the #Developer portal for a sneak peek. If you think you have a strong understanding of "how FPVs work" -- you might want to keep those flexible moving forward. One of the reasons I am hesitant to standardize training unnecessarily is that these systems are not static and their end use is not static. There's a lot more to do. Let the #builders build. 🦅 https://lnkd.in/g9PhGv-D #autonomy #uas #defense Defense Innovation Unit (DIU) Defense Advanced Research Projects Agency (DARPA) Office of Naval Research USSOCOM Technical Experimentation United States Air Force Air Force Research Laboratory Army Applications Laboratory Marine Innovation Unit Fifth Corps U.S. Indo-Pacific Command Naval Postgraduate School European Defense Tech Baltic Ghost Wing Center of Excellence GNIZDO Ltd #ukrainewillwin

57

3 Comments

Pretty cool update out of the DIU and AFRCO. A quantum inertial sensor just got cleared for launch on the X-37B. This is the first time DIU has teamed up with that program, and it’s the kind of hardware we’ve been waiting to see move out of labs and into mission-relevant environments. Props to the teams that worked on this. Building something that meets the bar for space isn’t easy. This kind of integration (non-trad hardware, traditional platforms, actual ops-focused use cases) is what it looks like when the system works. Curious to see how this shapes up for PNT and anomaly detection. Getting real data from orbit changes the game. Secured Quantum Services #Quantum #DIU #EmergingTech #MissionReady https://lnkd.in/e_TPndXg

16

1 Comment

Vital avionics systems functions which in some cases are deemed safety-critical are increasingly Software dependent. DO-178C considers software development techniques like Model-driven engineering but thankfully ARINC 653 or VxWorks 653 have made our software development lives a bit easier when certification time comes. Some other software engineering docs to consider are found in NASA's technical library which reflect IEEE and ACM software engineering practices from the past 40 yrs that have proven to be human spaceflight effective if we keep our units consistent throughout our calculations and do an adequate level of software testing. Whatever level of software and spacecraft testing that your organization currently does is typically insufficient given today's budget and schedule constraints so we have to develop consistent processes, models and tools to always ensure success. https://lnkd.in/eHjH-Pkp https://lnkd.in/edDYdZvA https://lnkd.in/e4Trkuub

1

The software engineering abilities of LLMs are accelerating at a phenomenal pace. Anthropic have recently released Opus 4.6. As a demonstration of the capabilities of Opus 4.6 Anthropic have today published an article and video on a Claude Code Agent Team building a GNU C compiler from scratch that can target multiple architectures (x86-64, ARM, RISC-V) with the end result passing the GCC Torture Test Suite. This did cost $20,000 in token API calls running 2000 concurrent Claude Code sessions. As the author of the article concludes, nobody expected to LLMs to be capable of doing anything close to this in 2026 and, quote, "we’re entering a new world which will require new strategies to navigate safely". I agree https://lnkd.in/erBmxD6m

26

3 Comments

This new “Right to Integrate” (R2I) initiative validates a major shift happening across defense tech… the future is not isolated systems operating in silos… it’s interoperable mission environments. At the Mission Partner Environment (MPE) Summit last month, Alleo demonstrated this capability in action… a unified interaction layer sitting above the stack… bringing together disparate systems, files, data sources, applications, maps, video, etc… connecting mission partners in a synchronized Common Operating Picture. #COPofCOPs We look forward to helping advance this initiative and supporting the next generation of interoperable mission environments.

21

I just read Nate B. Jones article on dark factories. Three engineers at StrongDM now run a software factory where no human writes or reviews code. Specs go in, shippable software comes out. Coming from systems engineering, we've lived in this world. Specifications, requirements, verification, validation - that's just the V-lifecycle in a different environment with a faster feedback loop. Nate talks about a 5 level framework for AI development - level 5 is where you're building software with specs alone, test cases as scenarios, verification that it meets the spec, validation that it meets real use. That's not a new concept to us. SE could play a real part in this space, we just have to strip back the process. The document-heavy approach exists because in aerospace the cost of failure is catastrophic - nobody dies if an app crashes. We're sitting on decades of structured thinking that maps directly onto this. Keep the core: define what the system must do, verify it actually does it. #systemsengineering #AI #automation #darktactory #buildinpublic

28

7 Comments

From the classroom to the cosmos! This year, we've reached thousands of students through $16 million in STEM investments, from FIRST® Robotics to the Mission Mars Challenge, inspiring tomorrow’s explorers. During World Space Week, we’re celebrating how STEM education and innovation fuel the mission of Living in Space — empowering the next generation to imagine, design and lead humanity’s journey beyond Earth. Read more: https://lmt.co/4o4vDGW

1

If we're adventuring into the (digital or "MBSE") Systems Engineering realm, we must not dismiss the many opportunities for ROI that equally prioritizes the warfighter when coupling digital sustainment requirements with the digital design development processes.... FULL Product Lifecycle Systems Engineering is already a thing... "THOMAS: From my perspective, it’s really all about system engineering, and system engineering is really all about total life cycle performance and support. System engineering hits on architectures, requirements, and mission needs when and where we need them. How do we define and decompose our systems into modules or components? How do we select products or solutions that fulfill needs? How do we verify and field those functions and capabilities? Additionally, how do we insert new technologies, and how do we do all that with a bounded trade space of cost, schedule, performance (i.e. delivery need dates)? We also have to consider those elements for today as well as the evolving threat environment that we’re dealt." https://lnkd.in/g3A5JtRM #mbse #digitalengineering #systemsengineering #digitalsustainment #testability #dsi #express #diagnostics #MOSA #readiness #productlifecycle #reliabilityengineering #test #lifecycleperformance #missioncapacity #missioncapable #phm #rtx #ngc #dft #digitalthread #digitaltwin #rams #fmeca #systemsafety #maintenance #aargm #aamram #navair #nswc #nawcad #nawcwd

8

Vision Guidence System for an Interceptor: How a 2-pixel error becomes a 50-meter miss. At SpearX, we talks about "replacing radar with vision." But in reality, Vision-based guidance isn't just about swapping sensors. It is a complex "dance" between the algorithm and the physics of flight. When I look at our architecture, the camera isn't just a passive sensor. It is the heartbeat of the control loop. How do we make "camera → steering command" work reliably? 🔹 The Eyes: Geometry over Pixels A camera gives us coordinates, not truth. Lenses distort; vibrations shake the horizon. Our job at #SpearX is to teach the CUAS system to understand real space. We filter the noise so every pixel matches a real direction. 🔹 Time is the Enemy The chain "Capture → Process → Act" takes time. While the AI thinks, a fast target moves. If we miss 2 frames, we are chasing the past. We don't just detect; we predict where the target will be when our actuators finally move. 🔹 Scale Matters A 1-pixel shift is small on a screen but huge in the sky. We use optical flow logic: if the target stays in one spot in the frame, our paths will cross. But cameras measure angles, while engines control linear motion. The math must bridge this gap. 🔹 Trusting the Data Real-world data is noisy. We use smart filters to smooth the signal. But here is the key: our algorithms estimate confidence. If the vision is blurry or unsure, the controller becomes cautious. It doesn't guess; it manages risk. How we integrate Vision and Dynamics at SpearX: 1.Remove Self-Motion: We subtract the drone's own movement to see the target's true path. 2.Predictive Control: We sync our clocks tightly. We aim for where the target will be, not where it is. 3️.Shared Uncertainty: The vision module tells the flight computer: "I see it, but I'm only 85% sure." The flight computer listens and adjusts. My opinions: In autonomous systems, there is no "just an image." A tiny pixel error, multiplied by speed and delay, causes a real miss. That is why we believe in co-Design. We train vision with flight physics in mind, and we build flight controls that understand vision errors. We build honest systems that tell us not just where the target is, but how sure they are. Question from me: What is your biggest challenge when fast software meets hard physics? Let's discuss below. #VisionBasedGuidance #CTO #ComputerVision #GuidanceSystems #UAV #AutonomousSystems #SystemsEngineering #AI #ControlSystems #Robotics #SpearX #DefenseIndustry #EdgeAI #CV #Seekers #UAV #DefenseTech #ML #Interceptor #CUAS

24

4 Comments

Proud to see Pacific Northwest National Laboratory collaborating with four other DOE national laboratories on Steel Thread, a National Nuclear Security Administration (NNSA)-funded effort to build SeisModal—a multimodal AI model trained on the world’s largest public earthquake dataset. The goal of Steel Thread is to create a model that can analyze a broad set of scientific data relevant for nuclear nonproliferation. Read the full article here: https://lnkd.in/gXYBjBHi

29

The Army’s IFPC-HEL program signals a clear inflection point in counter-drone and air defense strategy. As low-cost UAS and swarm threats outpace traditional interceptors, the military is moving decisively toward directed energy solutions that scale. High-energy lasers, like those being developed by Lockheed Martin, and high-power microwave systems, such as Epirus’ Leonidas, show two complementary paths forward. What’s enabling this shift is embedded computing. Real-time processing, sensor fusion, and software-defined control are now just as critical as the energy source itself, pushing the force toward modular, compute-centric architectures across air and missile defense. The open question is whether industry can scale these systems fast enough to meet the threat. #EmbeddedSystems #CounterUAS #DefenseTech

15

At Space Systems Analytics, we're training a Decision Transformer to control a 6DOF spacecraft. The foundation is Andrej Karpathy's minGPT framework. For anyone who hasn't dug into it, miniGPT is a compact, readable implementation of a GPT-style transformer. It's designed to be hackable. You can strip it down, retrain it, and repurpose it for problems that have nothing to do with language which is what we're doing here. Instead of generating text, we've adapted it to output forces and torques conditioned on spacecraft state: position, velocity, quaternion, angular rates. The model learns from offline trajectories generated by a PD controller with noise injection, and outputs a stochastic control policy, mean plus variance, so I get a per-step uncertainty estimate. Here's a clip from a microgravity hold scenario. The red point is the target. The blue line is the actual trajectory. It stays close, but you'll notice drift and persistent attitude oscillations. That gap is the point. It learned what the trajectory should look like, but not how to stabilize under disturbance. We're now working toward integrating this into core Flight System (cFS) to evaluate how learned controllers behave under real flight software constraints. Longer term, I'm interested in how we quantify and trust autonomous control systems in mission-critical environments. If you're working in GNC, autonomy, or flight software, I'd be interested in your perspective. I'm in Phase 1. https://lnkd.in/egGATYrb

28

#AIM2025 is officially underway We're kicking things off with two powerful themes driving AI in materials and manufacturing today: Applied ML for Manufacturing I & II, real-world case studies and industrial applications, and Development of Novel ML Methodologies I & II, the latest research breakthroughs in machine learning tailored for this domain. Sessions start at 9:00 AM, after the first all-congress plenary at 8:00 AM. Explore both tracks throughout the day. #TMSSpecialtyCongress

8

This thought piece from Michal-Anne Rogondino on the role of User Centered Design for new systems like Golden Dome is worth a read. Most systems designed for military users are complex and cumbersome. The USSF has been making great strides to standardize user interfaces with tools like the AstroUX design system. Design can not only make the system easier to learn but can reduce cognitive load and reduce errors--something that's vitally important if you're about to shoot down a missile (or if you're testing a missile alert system as happened in Hawaii in 2018...) https://lnkd.in/grekPXdP

17

5 Comments

Anduril Industries and Riverside Research are partnering to bring DARPA’s Formal Methods technology out of the lab and into the field by pioneering efforts to mathematically eliminate cyber vulnerabilities in software-defined defense systems. This collaboration strengthens our Lattice platform and broader product suite, ensuring mission-critical systems remain secure, reliable, and resilient against adversary cyber attacks. A major step forward in cyber-hardening the tactical edge. https://lnkd.in/g82F8tTr

40