Max Robotics — Global Service Robot Directory + US Certification

About

Umberto Michieli is a computer vision researcher whose work sits at the intersection of robotics, continual learning, and robust visual recognition — areas critical to the development of intelligent, real-world AI systems. His research addresses some of the most pressing challenges in deploying vision models on resource-constrained devices, such as home robots and smart appliances, where adaptability, efficiency, and reliability are paramount. Among his notable contributions, Michieli has advanced online continual learning frameworks that enable robots to recognize new object categories without catastrophic forgetting of prior knowledge — a fundamental hurdle in lifelong AI systems (8 citations). His work on point cloud semantic segmentation introduces self-regularizing hierarchical representations that enhance fine-grained 3D scene understanding for autonomous agents (7 citations). More recently, he developed Swiss DINO, an efficient framework for personalized on-device object search in robotic appliances (3 citations), and proposed FFT-based methods for improving model robustness against severely corrupted images (2 citations). Collectively, Michieli's research pushes the boundaries of practical, deployable vision intelligence — making him a promising contributor to the growing field of edge AI and embodied robotic perception.