For Arduino and Rasperry Pi
Artificial intelligence for edge devices
Researchers at Microsoft Research Labs are currently developing machine learning services for microcomputers such as Raspberry Pi or Arduino. The aim is to make artificial intelligence usable for edge devices. An initial prototype of the software is now available on GitHub.
"Using the software, we can equip clothing, household appliances, cars or medical devices with intelligence, for example," explains Peter Jaeger, Senior Director Developer Experience and Evangelism (DX), Microsoft Germany. "This solves a whole range of challenges in the Internet of Things, such as limited bandwidths and excessively long latency times. Energy consumption is also reduced, as communication with the data center is eliminated and the data remains in the devices and is processed there."
Diverse application scenarios
As far as specific application scenarios for machine learning are concerned, the Microsoft research team sees very different scenarios: from intelligent soil humidifiers in agriculture to brain implants that warn people of epileptic seizures. The Microsoft research team is also working on applications that simplify the development of AI solutions for developers - from collecting and processing data to training models for machine learning and the ability to roll out these applications on any device.
According to Microsoft, the simplified use of AI technologies is a further step towards democratizing AI and enabling as many people as possible around the world to participate. The new open source offering is intended to promote the standardization of protocols and interfaces for autonomous devices in the real world.










