Max Robotics — Global Service Robot Directory + US Certification

About

Alberto Castellini is a researcher whose work sits at the intersection of autonomous robotics, probabilistic planning, and artificial intelligence safety. His most significant contributions center on Partially Observable Monte Carlo Planning (POMCP), an advanced framework for decision-making under uncertainty in large-scale environments. Castellini has substantially extended this algorithm's capabilities, developing novel approaches for mobile robot navigation, active visual search in indoor environments, and risk-aware policy generation — work that has collectively garnered dozens of citations and established him as a notable voice in the autonomous systems community. A particularly impactful thread of his research addresses the critical challenge of safe AI behavior. His development of rule-based and risk-aware shielding mechanisms for POMCP policies represents a meaningful advance in ensuring that autonomous agents operate within acceptable boundaries, a concern of growing importance as robots are deployed in real-world settings. Complementing this, his work on anomaly detection using Hidden Markov Models contributes to long-term robot autonomy and fault resilience. Castellini has also contributed to the data science community through a publicly available aquatic drone sensor dataset, reflecting a commitment to open, reproducible research. His body of work demonstrates a coherent vision: making autonomous systems smarter, safer, and more explainable.