Lightning Talk: Debugging Edge AI on Zephyr and Lessons Learned
Lightningat Open Source Summit Japan + AI_dev: Open Source GenAI & ML Summit + Automotive Linux Summit 2025 · 2025-12-08..12-10·
summary
Built an acoustic and motion recognizer that runs in kilobytes not megabytes, with ROM ~107 KB and RAM ~30 KB, and successfully deployed and debugged models on tiny hardware. Uses Zephyr + TensorFlow Lite Micro and model building via Edge Impulse, flashed to Pico + SH board over USB. Live tests show idling, up/down and short “freaking” events classify reliably, and all sources and Zephyr build instructions are on GitHub.
AI-written summary · about
related talks
07mLightning Talk: Zephyr Portability with an AI Application on Very Different MCUs
11mLightning Talk: Accelerating Experimentation...
14mKeynote: Leveraging Zephyr and ML to Bring Smart Devices to Market, Faster
55mEdgeAI - AI in the Palm of Your Hand
54mSmarter Everything with ML & the IoT
31mKeynote: The next big thing in machine learning is tiny
59mBuilding models for TPU devices
21mUnlocking the potential of low-power embedded devices with Edge AI software