Photonic neural network technology developed in the LOLIPOP Project enables processing of digital visual data by modulating signals onto a 1550 nm optical carrier and performing optical dot-product operations, a fundamental building block for future optical AI and image classification processors.
The platform combines TriPleX neuromorphic PIC technology, LNOI-based high-bandwidth Mach–Zehnder modulation and hybrid integration approaches such as edge-coupling and flip-chip bonding, supporting high-speed optical signal processing and precise characterization. Photonic integration and assembly of the demonstrator module was completed by PHIX Photonics Assembly.
The Packaged Neural Network module is a variation of the PHIX RF Characterization Package. It is an adaptation of the PHIX Photonics Assembly standard RF Characterization Package, where the RF section is a special customization.
This work contributes to scalable, high-performance photonic AI and next-generation optical computing systems.