Neuromorphic Sensing and Processing

Neuromorphic engineering is to mimic, as closely as possible, the way biological neural systems capture and process information. Neuromorphic artificial systems have been shown to reduce latency, energy consumption, transmission bandwidth, and computational processing costs.

As for the capture ( sensing ) part, ITI is working with commercial neuromorphic vision ( silicon retinas ) and hearing sensors .

In the processing part, we are working with commercial neuromorphic processors to implement Spiking Neural Networks ( SNNs ) within them.

Within ITI, the overall objective of this line of R&D is to research and develop new AI models based on neuromorphic principles , and to use all the results to transfer them to new applications. 

The research focus is primarily on the design and training of SNNs and the possibilities of interconnecting different types of neuromorphic hardware to integrate complete systems.

Some of the possible applications are autonomous driving/navigation of different types of vehicles (drones, robots, cars, etc.) or the detection of anomalies in industrial environments .

Current developments

·       High-speed object tracking ( neuromorphic vision )

·       Fluids ( neuromorphic vision )

·       Sound classification ( neuromorphic audio )