Zi Cong Guo

University of Toronto, Robotics Research (United States)

Papers

Total Citations

H-Index

About

Zi Cong Guo is pioneering the intersection of Koopman operator theory and robotics estimation, fundamentally rethinking how autonomous systems understand their environments. His flagship work introduces the Koopman State Estimator (KoopSE), a groundbreaking framework that achieves model-free batch state estimation for control-affine systems without linearization assumptions or problem-specific feature engineering—a paradigm shift from traditional approaches. This work, with 12 citations, demonstrates that inference costs remain independent of training data size, offering unprecedented scalability. Guo extended this framework to data-driven batch localization and SLAM, showing how lifting functions can render both process and measurement models bilinear in high-dimensional spaces, enabling simultaneous localization and mapping without explicit system models. Most recently, he developed closed-form expressions for marginalizing and conditioning Gaussians onto linear approximations of smooth manifolds, providing elegant mathematical tools for robotics applications on non-Euclidean spaces. His research bridges rigorous mathematical theory with practical robotics challenges, offering computationally efficient solutions that bypass traditional linearization bottlenecks. Guo's work is particularly impactful for researchers in SLAM, state estimation, and nonlinear control, where his Koopman-based methods promise to unlock new capabilities for autonomous navigation in complex, unstructured environments.