Ekkono provides a small footprint, platform-agnostic, all-software Edge Machine Learning solution for rapid development and deployment of smart, self-learning, and predictive features that run onboard things (IoT) from product OEMs in different verticals, like manufacturing, automotive, energy, building automation, MedTech, etc.

Besides being able to scale down to constrained platforms (performance benchmarks on MCUs, Cortex-M0+), which enables you to utilize Ekkono across all your products, from large to small; Ekkono also does machine learning, not just inference, at the edge. This incremental learning enables individual learning of normal behavior per device, early detection of deviations, instant insights instead of waiting while collecting data for a year, insights that get better and better over time, training on real-time, high-frequency sensor data instead of blunt averages sent to the cloud, re-training to new conditions, and improved data integrity.

The core of the product is a C++ or C library, surrounded by an API with bindings to Python and C#, and a comprehensive SDK with AutoML functionality, integration, and optimization tools. Ekkono has, together with Infineon, ensured optimal performance on Infineon’s platforms.

Ekkono Edge Intelligence:

Ekkono offers comprehensive edge machine learning (ML) software. The core is a C++ or C library. Only a few lines of code are required to call Ekkono’s library in deployment. The library supports multiple ML algorithms like regression trees, random forests, and neural networks. Ekkono enables pipelining, inference, training, anomaly/change detection, sensitivity analysis, and conformal prediction to run at the edge in real-time and in memory.

The Ekkono SDK includes AutoML tools, and Studio provides pre-defined code snippets. The run-time – provided as source code for compilation on your target platform – is separated from the ML model for seamless migration between Python, C, and C++, and model hot-swapping in operation.

Virtual Aperture Imaging Software:

• Traditional radars have poor angular resolution and limited FOVs because traditional designs require more antennas for higher resolution. Additional antennas increase cost, size, and power exponentially, limiting what is commercially feasible

• Oculii’s Virtual Aperture Imaging Software uses an adaptive AI-driven waveform to increase the angular resolution without requiring additional antennas.

• Oculii's Virtual Aperture Imaging Software can be embedded into Infineon’s AURIX processing platform to enhances the angular resolution of any radar hardware platform with any number of MMIC frontends.

Automotive See-With-Sound (SWS):

Cameras, lidar, and radar can’t see around corners, and they can’t see emergency vehicles that are far away but fast approaching. With the addition of obstructed view scenarios into the safety standards of Euro NCAP and other international testing bodies, finding new ways to detect threats is critical.

Reality AI’s Automotive See-With-Sound (SWS) system detects and localizes threats that LOS sensors can’t see. Adding Automotive SWS to ADAS and AV systems makes vehicles even safer by using sound to detect and position a wide array of common threats to passenger safety. SWS uses XENSIV™ MEMS microphones(IM67D130A), the AURIX™ Microcontroller (TC387), and unique detection and localization software to give cars the ability to see around corners and through visual obstacles. This solution can detect and localize emergency vehicles, cars, trucks, motorcycles, bicycles, and joggers.

With the powerful combination of Infineon hardware, detection, and localization provided from our AI models, and our AI-engineered firmware, Reality AI’s solution gives passengers a new level of protection.

Availability

• The XENSIV MEMS microphone IM67D130A can be ordered now in a PG-LLGA-5-4 package. More information is available at www.infineon.com/mems-microphones. The demonstrator will be presented at Infineon’s Virtual Sensor Experience.
• Click here to have a look on the full AURIX™ TC3xx scalable microcontroller family 

 Related links

Click here and take a look at the press release

Teraki has successfully implemented its embedded client for Intelligent Signal Processing on the latest generation of Infineon’s AURIX™ microcontrollers (TC3xx), one of the most commonly used real-time controllers in the automotive industry. A demonstration by both companies shows the ability to simultaneously process 50 signals and reduce the data by 95% at low latency at a consistently low percentage of CPU capacity.