Название: Embedded AI: Intelligence at the Deep Edge
Автор: David Such
Издательство: No Starch Press
Год: 2027
Страниц: 587
Язык: английский
Формат: True/Retail PDF, True/Retail EPUB
Размер: 82.9 MB

A project-driven guide to designing, training, and deploying Artificial Intelligence directly on embedded hardware, showing how to build intelligent, autonomous systems under real-world constraints.

You already know how to build embedded systems. Now it’s time to make them intelligent.

Adding AI to an embedded device takes more than training a model. You have to choose the right hardware, collect and prepare data, deploy models to resource-constrained devices, and integrate everything into a system that performs reliably.

Drawing on more than 30 years of embedded engineering experience, David Such takes you through the complete engineering process. You’ll work through more than 25 hands-on projects (complete with downloadable source code, schematics, PCB designs, and datasets); no Machine Learning experience required.

The book takes a project-based approach to embedded AI, using hands-on, real-world examples to teach key concepts. Each chapter is anchored around practical projects that demonstrate the theory—such as detecting orientation with a gyroscope, predicting battery health, or building a voice assistant. Designing and deploying complete systems will give you technical knowledge while also developing your intuition and confidence. Both are essential for applying AI in embedded contexts. These projects are designed to develop a deep, layered understanding of the material—something a purely theoretical treatment can’t replicate. Because projects deliver measurable outcomes, they demand rigor. You can’t hand-wave a wiring diagram, ignore timing constraints, or skip model deployment steps; the hardware won’t let you. This necessity forces clarity of thought and careful execution, making project-based learning a powerful way to internalize complex material and expose gaps in understanding that theory alone might miss. Embedded AI is inherently multidisciplinary. A single project might span signal processing, machine learning, hardware design, and firmware. In the embedded world, hardware and software are tightly coupled, so this book treats them that way.

You’ll build:

A wake-word detector that responds to your voice
A real-time AI noise suppressor
An AI-powered MIDI synthesizer that composes music
A battery monitor that collects its own training data
A person detector that runs a neural network on a camera board

Whether you’re an embedded developer adding AI to your products, a Machine Learning practitioner moving onto embedded hardware, or a maker ready to move beyond beginner projects, Embedded AI teaches you the engineering decisions behind every design. When the breadboard is flaky, the sensor data is noisy, or the tensor arena is too small, you’ll know how to fix it—and why.

Prerequisites:
Most projects require an Arduino UNO or Raspberry Pi Pico; a few use specialized boards. You’ll also need to download some free software, including Python with TensorFlow, Arduino IDE, and Raspberry Pi Pico SDK.

Contents: