🤝 Top Collaborators
🤓 Latest Submissions
ZeroCrash AI: Industrial Safety Bridge
(Problem Statement) In precision manufacturing and CNC machining, a single typographical error in G-code—such as a misplaced decimal in a Z-axis plunge or an excessive feed rate—can result in catastrophic machine crashes. Traditional controllers execute instructions blindly, assuming the code is flawless. When errors occur, the result is thousands of dollars in broken tools, ruined materials, and potential safety hazards for the operator. Currently, safety relies entirely on human vigilance and slow, offline simulation software. (The Solution) Enter ZeroCrash AI. Inspired by rigorous aviation-style safety pre-checks, we built an intelligent, decentralized safety bridge between the operator's intent and the physical machine's motion. ZeroCrash AI integrates seamlessly into the execution ecosystem (like LinuxCNC) to act as an autonomous, real-time safety inspector. (Target Users & Interaction) Designed for CNC machinists, mechanical engineers, and factory operators, the system requires zero changes to their existing workflow. The operator simply loads their job file into the machine interface as usual. Behind the scenes, ZeroCrash AI instantly intercepts the file. Before the spindle is ever allowed to turn, the system sends the movement vectors to IBM Bob 2.0 for a high-speed safety audit. If the code is safe, the machine executes the cut with near-zero latency. If IBM Bob detects a hazardous anomaly (e.g., unsafe feed rates, dangerous toolpaths, or missing stops), the job is immediately rejected, and the machine stays idle, alerting the operator to the specific danger. (What Makes it Unique) Unlike traditional post-processors or offline simulators, ZeroCrash AI is an active, real-time gatekeeper. It leverages a ZeroMQ messaging protocol to ensure lightweight, lag-free communication. By shifting the paradigm from "repairing after a crash" to "preventing the crash entirely," we are bringing an unprecedented standard of AI safety to the factory floor.
27 Sep 2026
INGEK - Universal Fleet Control
By switching from a reactive post-crash investigative paradigm to proactive live mitigation, INGEK (Universal Fleet Control), a next-generation intelligent ground station, aims to transform international aviation safety. For more than 50 years, crucial flight information was kept hidden inside actual black boxes located far below the surface of the ocean, and during serious sensor abnormalities like pitot icing or engine overheating, ground control operated in a blind manner. By functioning as a living "Digital Twin" that replicates real-time aircraft vital signals from the skies directly to the ground control center, INGEK closes this information gap. Our software, which is housed inside a secure Docker container, uses Fireworks AI to use the huge brute-force acceleration of AMD GPUs to process and synthesize millions of complicated raw telemetry data streams in less than three seconds. The AMD-powered Llama 3.1 8B AI immediately analyzes the telemetry package and creates enterprise-grade Emergency Action Protocols for the ground crew in the event of a critical failure. INGEK prevents catastrophic aircraft losses and reduces airline fleet management operating expenses by up to 80% by providing dual-engine flexibility between real-time data streaming and high-precision historical data offset retrieval.
13 Jul 2026
