Coredump Sessions · 25 Feb 2025 · 58m · Episode 5
Alex Samuelson, CTO and co-founder of Imagimob, joins Memfault founders Francois Baldassari and Chris Coleman to discuss the practical implementation of Edge AI on embedded devices. They highlight how hardware accelerators like the Cortex-M55 and U55 have enabled GPU-like performance on microcontrollers, allowing for real-time inference in applications such as welding quality monitoring and fall detection. The conversation emphasizes that data collection, labeling, and memory constraints are the primary bottlenecks, requiring device makers to build data acquisition firmware early in the development process. Finally, they discuss the shift toward 'AI as a Service' business models, where hardware companies generate recurring revenue by updating models over the device's lifecycle.
Where this conversation describes AI being used to build or deliver a product. Part of the AI in practice tracker.
Engineering & simulationImagimov platform6 months of battery life
A tiny neural network was deployed on a battery-powered watch to detect falls in the elderly, enabling the device to function for months on a non-rechargeable battery.
“We were doing full detection. really early for for the elderly. … we could do it with a JI. Also with a tiny network, on a battery powered watch … with a non-rechargeable battery for the ability that would last for months.”
Speaker 1 · Hear it at 15:22
Engineering & simulationInfineon PSOC 6 AI
Infineon released SDK updates for the PSOC 6 AI chip to simplify the training and deployment of machine learning models on their microcontrollers.
“I've seen Infineon, for example, has a number of chips and SDK updates to make it very easy to train and deploy these models. The PSAC 6 AI. … was one of the big announcements.”
Speaker 2 · Hear it at 6:42
Engineering & simulationImagimov Deep Craft Studio
The tool allows developers to generate consistent APIs for inference engines, abstracting away the complexity of choosing between proprietary backends or TensorFlow Lite Micro.
“If you're using a craft studio, we're making sure that the API is consistent, so you don't have to care about all of those details, and you can bring your own model into that and leverage.”
Speaker 1 · Hear it at 24:41
early-stage companies
Other shows on the same ground, closest first.
Renesas Electronics EVP Silas Chitapetti explains how the convergence of AI, IoT, and 5G creates 'intelligent' devices through AIoT, shifting processing from the cloud to the edge. He details Renesas' strategy of embedding AI acceleration in microcontrollers and microprocessors, supported by the acquisition of Reality AI to handle industrial machine learning workloads. Chitapetti highlights specific use cases like predictive maintenance using current sensing, vision-based authentication, and robotic food preparation, noting that these edge AI applications reduce hardware costs and improve response times.
Nitin Dahad, Francoise Pelle, and Sally Ward Foxton of EE Times recap Embedded World 2023, highlighting the shift toward edge AI, tiny ML, and software-defined vehicles. They discuss specific implementations from companies like Renaissance, STMicroelectronics, and Silicon Labs, focusing on running AI models on Cortex-M85 and STM32 microcontrollers with low power consumption. The episode also covers security initiatives from Kudosip and Trust In Software, as well as neuromorphic computing demos from InnerTense and Brain Chip.
Nitin Dahad interviews Paul Williamson of Arm, Daniel Cooley of Silicon Labs, and Fiona Tracy of Analog Devices at Embedded World 2024. Williamson discusses Arm's new Ethos U85 AI accelerator and the shift toward transformer networks on edge devices. Cooley details Silicon Labs' XG26 chip, emphasizing low-power inference on battery and Matter connectivity. Tracy outlines ADI's focus on sustainable, software-configurable factories and real-time awareness through sensor fusion and edge AI.
Kaushal Vora and Mo Dogar from Renesas discuss the convergence of AI and IoT (AIoT) at the edge, distinguishing between resource-intensive generative AI and predictive AI suited for constrained endpoint devices. They highlight Renesas' acquisition of Reality AI and its AutoML toolchain, which allows embedded engineers to optimize machine learning models for microcontrollers without adding hardware sensors, as demonstrated in washer and battery management use cases. The guests explain how decentralizing intelligence reduces latency, power consumption, and security risks by performing inference locally on devices like the RH850 microcontroller.
Justin Schneck, co-founder and CPO of Peridio, discusses the evolution of AI from cloud to edge, introducing Peridio's upcoming 'Avocado' OS designed to simplify embedded product development. He and host Ryan Prosser of Very explore the spectrum from cloud AI to embedded AI, noting how specialized silicon and miniaturized models are enabling practical applications like Garmin's radar-enabled bike lights. Schneck argues that while cloud AI handles massive compute, the real value lies in robust, self-initializing embedded systems that reduce hardware complexity for developers.
The summary, chapters and topics are generated from the episode’s audio by this directory, not written by Coredump Sessions. The episode itself belongs to its publisher.