Research Area

Our brain is an amazing low-power, always-on machine with a remarkable ability to integrate and segregate signals across various timescales robustly. My research focuses on understanding these mechanisms and leveraging them to build and exploit bio-inspired neuromorphic systems for processing and combining real-world signals, addressing practical challenges through edge devices.

Academic Career

I hold a degree in Theoretical Physics from La Sapienza University in Rome, where I specialized in Complex Systems Dynamics. My early research at the National Institute of Nuclear Physics in Rome explored the role of sleep in learning and cognition using bio-inspired spiking neural network models. As a Postdoc in Giacomo Indiveri's Lab at the Institute of Neuroinformatics in Zurich, I worked on exploiting low-power mixed-signal neuromorphic chips for always-on edge systems. Currently, I am part of the Digital Society Initiative (DSI) at the University of Zurich, focusing on applying neuromorphic technologies in precision farming to enhance and optimize agriculture and plant monitoring.

Projects

Within the scope of my DSI Bridge Postdoc Fellowship, I am conducting my research project «Neuromorphic technologies for precision farming». The tremendous success of Artificial Intelligence (AI) is providing important solutions to societal challenges in multiple domains. However, modern agricultural production relies on monitoring the status of the environment by observing and measuring variables continuously over extended periods of time, often in remote locations. In this respect, conventional AI approaches are not always ideal, as they have high energy costs and often require access to the network, in order to operate. Neuromorphic hardware offers a promising solution to this challenge, as it enables power-efficient always-on sensing and computing, using local stand-alone sensory-processing systems that do not need to transmit high amounts of data to the cloud. In this project, we propose to integrate neuromorphic computing technology into precision farming. Ultra-low-power mixed-signal VLSI neuromorphic devices will be employed for on-sensor signal processing, enabling local operations such as noise filtering, trend measurement, event detection, and multi-modal classification. The aim is to  reduce the data that needs to be transmitted to the central computing unit, and open opportunities for new precision farming strategies and applications that are currently not realizable with conventional AI technologies.