Cartesiam IDE adds edge anomaly classification on Arm Cortex-M MCUs

Cartesiam has introduced a new version of its integrated development environment (IDE) which it said is the first to enable anomaly classification directly on all Arm Cortex microcontrollers (MCUs). It has also introduced a new-web based platform for users to download real datasets of representative use cases, and announced a partnership with Bosch Connected Devices and Solutions to extend its IoT product line with Cartesiam’s IDE.

Having previously introduced earlier this year an IDE for developers to create artificial intelligence (AI) training and inference applications on microcontrollers, the company has now announced the availability of NanoEdge AI Studio V2, which in addition to simplifying creation of machine learning (ML) and inference, now adds classification libraries for direct implementation on Arm Cortex-M MCUs.

Cartesiam said this new IDE has a superior approach to anomaly detection and classification. This is because the model is trained in the microcontroller, which means anomaly detection wakes up the classifier for characterization, telling the system exactly what’s wrong, not just that there’s a generic problem’ this is the key to giving users the intelligence needed to make more informed decisions.

Cartesiam NanoEdge AI screenshot1
In addition to indicating that there is a problem, Cartesiam’s new version IDE can help tell the system exactly what’s wrong, providing an additional level of intelligence needed to make more informed decisions. (Image: Cartesiam)

Joël Rubino, CEO and co-founder of Cartesiam, explained to, “Our solution has been designed from day one within the box of a microcontroller. We re-develop, from the algebra, all machine learning (ML) and signal processing algorithms to run natively inside an MCU. Other solutions on the market are ‘scaled down’ solutions from a framework designed to run on servers with unlimited computing power, memory, datasets and so on, to fit into a microcontroller and therefore our libraries are by far much more optimized versus the competition – such as Google TensorFlow and other AI software solutions running in the cloud. We usually fit into 4Kb of RAM in a typical configuration, and most of the time below 1Kb.”

Optimized for Arm Cortex-M MCUs, Cartesiam said its IDE does not require the expertise of data scientists and signal processing engineers, since it is an intuitive desktop tool that lets embedded developers focus on solving business problems rather than on selecting algorithms. It enables rapid learning at the edge, performing iterative learning in 30msecs in an Arm Cortex-M4 80MHz to deliver intelligence quickly.

The company said thousands of commercially available industrial IoT (IIoT) embedded devices are already in production with NanoEdge AI Studio V1 for anomaly detection. With the addition of classification libraries to NanoEdge AI Studio V2, developers can now more easily go beyond anomaly detection to qualify problems directly in endpoints.

“Cartesiam makes tools for embedded developers, offering an intuitive push-button approach that requires no background in data science, opening AI to the billions of resource-constrained embedded devices built with Arm Cortex-M MCUs,” Rubino, commented. “We initially designed NanoEdge AI Studio to meet demand from our customers in predictive maintenance, who, having accumulated data on the use of their equipment, asked us to help them easily qualify their events as well as to anticipate them. The new version of our IDE allows those customers — and any other embedded designer — to effortlessly develop a classification library without the usual challenges associated with signal processing and machine learning skills. This dramatically reduces costs and speeds time to market.”

He added, “Our solution runs on a PC. No cloud connection or cost required. Many companies, especially Europeans, are sceptical about sending their data to the cloud (due to data privacy concerns), and the hidden cloud computing cost.”

Sample datasets at new web-based platform, Bosch IoT partnership
Cartesiam also introduced a ‘use case explorer’ at, a new web-based platform. Users can download real datasets and try the NanoEdge AI Studio IDE on representative use cases, such as ventilator obstruction detection, breast cancer detection, vacuum-bag volume detection, and others. The company said it will continuously enhance the portal with additional datasets. use cases explorer
Users can download real datasets and try the NanoEdge AI Studio IDE on representative use cases (Image: Cartesiam)

Simultaneously with the launch of its new IDE and web platform, Bosch Connected Devices and Solutions is adding Cartesiam’s NanoEdge AI Studio to extend its existing IoT product line, the cross domain development kit, or XDK.

Bosch XDK
The Bosch XDK combines a wide array of MEMS sensors with a microcontroller, and now plans to use Cartesiam’s IDE to process data for anomaly detection and classification on one or more sensors (Image: Bosch)

Ando Feyh, head of technical responsibility, Bosch Connected Devices and Solutions, said, “With its range of eight sensors, the XDK platform lets designers monitor, control and analyze processes remotely via Bluetooth or Wi-Fi, enabling our customers to quickly create more intelligent connected machines. NanoEdge AI Studio V2 increases the XDK’s unique functionality, providing the ability to process data for anomaly detection and classification for one or more sensors. Given this, we plan to use Cartesiam’s platform in a wide range of internal and external projects, and are closely working together with Cartesiam on a NanoEdge AI Studio integration with our XDK.”

Related Contents:

For more Embedded, subscribe to Embedded’s weekly email newsletter.

The post Cartesiam IDE adds edge anomaly classification on Arm Cortex-M MCUs appeared first on

Original article: Cartesiam IDE adds edge anomaly classification on Arm Cortex-M MCUs
Author: Nitin Dahad