Components
TinyML: Running AI on Microcontrollers in 2026
How small neural networks run on boards like the ESP32-S3 for voice, vision and sensor tasks.
· 3 min read
Key takeaways
- TinyML runs models in kilobytes of RAM.
- ESP32-S3 has vector instructions for AI.
- Use cases: wake words, gesture and anomaly detection.
What TinyML is
Quantised neural networks run directly on low-power chips without the cloud, improving privacy and latency.
Best boards
The ESP32-S3 with PSRAM handles small vision and audio models.
Sensor fusion
Combine distance data from a LiDAR sensor with a small model to classify objects or gestures.
Real-time control
An STM32 Cortex-M4 with FPU runs compact models alongside motor control.
Frequently asked questions
Products mentioned in this article
Keep reading
Electronics
Noise-Cancelling Headphones Buying Guide: What Actually Matters
Hybrid ANC, codecs, battery life and multipoint explained — how to pick wireless noise-cancelling headphones without overpaying.
7 min read
Components
ESP32-S3 vs STM32: Which Microcontroller Should You Use?
A practical comparison of ESP32-S3 and STM32F411 boards for IoT, robotics and maker projects — connectivity, power, tooling and price.
9 min read
Carbon Fiber & Materials
Carbon Fiber Sheets Explained: 3K Twill, Thickness and CNC Tips
What 3K twill means, how to choose carbon fiber sheet thickness, and how to cut and drill carbon fiber safely for drones, RC and builds.
8 min read