A. Papalia - Assistant Professor of Naval Architecture & Marine Engineering, University of Michigan

Описание: “Algorithmic Foundations for Autonomous Ocean Observation”

Abstract: Marine environments remain one of the most challenging domains for autonomous navigation—sensing limitations and the absence of GPS make reliable localization especially difficult. Yet robust autonomy in these settings is essential for sustained ocean observation. In practice, the limitations of standard approaches (e.g., nonlinear least-squares) pose a major barrier to scalable ocean understanding.

This talk presents an alternative to conventional localization: certifiable estimation, which computes globally optimal solutions to non-convex problems along with certificates of correctness. I will focus on recent advances in range-aided SLAM, particularly CORA (Certifiably Optimal Range-Aided SLAM), and show how its structure—grounded in semidefinite relaxations—enables global optimization faster than the widely used toolkits which only perform local optimization.

Time permitting, I will also discuss our work on exploiting separable structure — which can be identified in range-aided SLAM problems — to allow for substantial computational advances in a wide range of inference and estimation problems.

Bio: Alan Papalia is an assistant professor in Naval Architecture & Marine Engineering at the University of Michigan, where he runs the Robotic Exploration Lab. He received his PhD in Ocean Engineering from MIT and the Woods Hole Oceanographic Institution in 2025. His research focuses on algorithms to enable robots to explore, perceive, and map unstructured environments. Specific research interests include simultaneous localization and mapping (SLAM), marine robotics, and certifiable perception. He is a recipient of the IEEE Transactions on Robotics King-Sun Fu Best Paper Award and was selected as a 2025 Robotics: Science and Systems Pioneer.