Max Robotics — Global Service Robot Directory + US Certification

About

Gabriele Graffieti is a leading researcher in continual learning for real-world robotic and embedded systems. His work focuses on enabling deep neural networks to learn incrementally from streaming data without catastrophic forgetting—a critical challenge for autonomous agents operating in dynamic environments. Graffieti’s most influential contribution is the development of **latent replay for real-time continual learning**, a technique that efficiently replays compressed representations from previous tasks to maintain performance while respecting the severe memory and compute constraints of edge devices. This work, published in 2020, has already garnered **138 citations**, underscoring its impact on the field. He was also a key participant and contributor to the **IROS 2019 Lifelong Robotic Vision Challenge**, where his methods were tested against over 150 teams on the OpenLORIS benchmark, advancing the state of the art in lifelong object recognition. Graffieti’s research bridges the gap between theoretical continual learning and practical deployment on robotic platforms, making him a notable figure in the push toward truly autonomous, learning-while-operating machines.