The NanoEdge AI Studio is an intuitive software tool that has been designed to allow system designers using Arm’s low-power, low-cost microcontrollers to easily and inexpensively integrate machine learning directly into everyday objects (industrial machines, IoT, automotive and household appliances).
Until now, implementing AI in embedded devices has proved to be a laborious, difficult and expensive process requiring the expertise of data scientists spending months or years of development time, and access to complex and extensive data sets that are difficult to source.
“Cartesiam’s NanoEdge AI Studio offers a completely different approach, with a cost- and time-efficient and self-learning AI,” explained Marc Dupaquier, general manager and co-founder, Cartesiam. “It allows any embedded designer to develop application-specific machine learning libraries quickly and run the program inside the microcontroller right where the signal becomes data. It’s the only solution that can run both machine learning and inference on the microcontroller.”
The NanoEdge AI Studio has been tested and deployed by a number of European and US companies and works by transforming passive sensors into autonomous agents capable of self-monitoring.
“Imagine an air conditioner that detects when its filter needs to be changed or an escalator activating its own preventive maintenance,” explained Dupaquier. “As far as security and privacy are concerned, learning an initial state locally reduces data exchanges over the network and prevents risk of falsification or intrusion. With our customers' creativity and innovation, there will be no limits to the development of inventive solutions based on NanoEdge AI Studio.”
NanoEdge AI Studio looks to remove traditional AI barriers and is intended for companies that either do not have expert resources in machine learning or that want to provide their data scientists with a complementary tool for embedded environments.
NanoEdge AI Studio Technical Facts
- Runs autonomously on the developer's workstation under Windows or Linux. Thus, no data is transmitted outside the customer’s environment.
- Will automatically test, optimise and calculate the best algorithmic combination among more than 500 million possible combinations, after the developer has described the targeted environment
- Provides the selected algorithm as a C library that is easily embeddable in the microcontroller
- Generates libraries that require only 4K to 16K of RAM, making them the most optimized AI algorithms in the industry
- Enables the execution of unsupervised learning, inference and prediction on the device edge, opening new classes of small, low-power, low-cost devices to AI for the first time
