Embedded Artificial Intelligence for IoT Applications Using the MAX78000
Martina Balbi, Lance Doherty, Thomas Watteyne · IEEE Access · 2025
Recent advances in embedded AI and IoT have revolutionized the development of intelligent edge devices. This work provides a tutorial on developing and deploying AI models on the MAX78000, a low-power microcontroller designed specifically for AI applications. Starting with the foundational understanding of neural networks and machine learning, this tutorial explores the architecture and capabilities of the MAX78000, which integrates a CNN accelerator with an ARM Cortex-M4 core. We give practical guidance on creating, training, and quantizing AI models, detailing essential tools, frameworks, and the deployment process. Real-world examples illustrate the versatility of AI microcontrollers and their performance in various IoT applications. Emphasizing the importance of accessible development tools, this tutorial aims to increase awareness within the IoT community about low-power accelerators. This enables developers to create efficient, real-time AI solutions, highlighting the transformative potential of embedded AI in IoT.
Cited by 4 works. doi:10.1109/access.2025.3546557
Authors
References
- Adam: A Method for Stochastic Optimization
- Deep learning
- Review of deep learning: concepts, CNN architectures, challenges, applications, future directions
- Machine Learning for Medical Imaging
- Equal numbers of neuronal and nonneuronal cells make the human brain an isometrically scaled‐up primate brain
- An Overview of Overfitting and its Solutions
- Progress in Speech Synthesis
- Activation Functions: Comparison of trends in Practice and Research for Deep Learning
- Machine learning algorithm validation with a limited sample size
- Using GPUs for machine learning algorithms